Vm-dev March 2016

vm-dev@lists.squeakfoundation.org

33 participants
139 discussions

Re: [Vm-dev] VM Maker: VMMaker.oscog-nice.1754.mcz
by Nicolas Cellier 31 Mar '16

31 Mar '16

Note that I have these "changes" working for 5 months now, so they should be reasonnably stable I'm not particularly happy with the proliferation of type hint and some of them might well be unecessary, but I don't believe in chirurgical patch: too fragile, I prefer uniformity. 2016-03-31 22:06 GMT+02:00 <commits(a)source.squeak.org>: > > Nicolas Cellier uploaded a new version of VMMaker to project VM Maker: > http://source.squeak.org/VMMaker/VMMaker.oscog-nice.1754.mcz > > ==================== Summary ==================== > > Name: VMMaker.oscog-nice.1754 > Author: nice > Time: 31 March 2016, 10:04:44.274 pm > UUID: a0138c60-3e62-446e-a1d8-b00b7bb6c92b > Ancestors: VMMaker.oscog-eem.1753 > > BitBltSimulation is operating on bits of 32bits words, therefore it's > better to declare its operans as 'unsigned int' rather than sqInt. > > On 32bits VM, this should not change anything, but on 64bits spur, it > makes this snippet work: > > | wideString source pos blt expectedWideString | > source := #[1 64 255 14 1 64 48 251]. > expectedWideString := WideString fromByteArray: source. > wideString := WideString new: source size // 4. > pos := 0. > blt := (BitBlt > toForm: (Form new hackBits: wideString)) > sourceForm: (Form new hackBits: source). > blt > combinationRule: Form over; > sourceX: 0; > sourceY: pos // 4; > height: wideString byteSize // 4; > width: 4; > destX: 0; > destY: 0; > copyBits. > wideString restoreEndianness. > self assert: wideString = expectedWideString > > Hence it fix loading/diffing MCZ with wide character. > > =============== Diff against VMMaker.oscog-eem.1753 =============== > > Item was changed: > ----- Method: BitBltSimulation>>addWord:with: (in category 'combination > rules') ----- > addWord: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord + destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>alphaBlend:with: (in category > 'combination rules') ----- > alphaBlend: sourceWord with: destinationWord > "Blend sourceWord with destinationWord, assuming both are 32-bit > pixels. > The source is assumed to have 255*alpha in the high 8 bits of each > pixel, > while the high 8 bits of the destinationWord will be ignored. > The blend produced is alpha*source + (1-alpha)*dest, with > the computation being performed independently on each color > component. The high byte of the result will be 0." > | alpha unAlpha result blendRB blendAG | > <inline: false> > + <returnTypeC: 'unsigned int'> > - <return: 'unsigned int'> > <var: #sourceWord type: 'unsigned int'> > <var: #destinationWord type: 'unsigned int'> > <var: #blendRB type: 'unsigned int'> > <var: #blendAG type: 'unsigned int'> > <var: #result type: 'unsigned int'> > <var: #alpha type: 'unsigned int'> > <var: #unAlpha type: 'unsigned int'> > alpha := sourceWord >> 24. "High 8 bits of source pixel" > alpha = 0 ifTrue: [ ^ destinationWord ]. > alpha = 255 ifTrue: [ ^ sourceWord ]. > unAlpha := 255 - alpha. > > blendRB := ((sourceWord bitAnd: 16rFF00FF) * alpha) + > ((destinationWord bitAnd: 16rFF00FF) * > unAlpha) > + 16rFF00FF. "blend red and blue" > > blendAG := (((sourceWord>> 8 bitOr: 16rFF0000) bitAnd: 16rFF00FF) > * alpha) + > ((destinationWord>>8 bitAnd: 16rFF00FF) * > unAlpha) > + 16rFF00FF. "blend alpha and green" > > blendRB := blendRB + (blendRB - 16r10001 >> 8 bitAnd: 16rFF00FF) > >> 8 bitAnd: 16rFF00FF. "divide by 255" > blendAG := blendAG + (blendAG - 16r10001 >> 8 bitAnd: 16rFF00FF) > >> 8 bitAnd: 16rFF00FF. > result := blendRB bitOr: blendAG<<8. > ^ result > ! > > Item was changed: > ----- Method: BitBltSimulation>>alphaBlendConst:with: (in category > 'combination rules') ----- > alphaBlendConst: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > > ^ self alphaBlendConst: sourceWord with: destinationWord > paintMode: false! > > Item was changed: > ----- Method: BitBltSimulation>>alphaBlendConst:with:paintMode: (in > category 'combination rules') ----- > alphaBlendConst: sourceWord with: destinationWord paintMode: paintMode > "Blend sourceWord with destinationWord using a constant alpha. > Alpha is encoded as 0 meaning 0.0, and 255 meaning 1.0. > The blend produced is alpha*source + (1.0-alpha)*dest, with the > computation being performed independently on each color component. > This function could eventually blend into any depth destination, > using the same color averaging and mapping as warpBlt. > paintMode = true means do nothing if the source pixel value is > zero." > > "This first implementation works with dest depths of 16 and 32 > bits only. > Normal color mapping will allow sources of lower depths in this > case, > and results can be mapped directly by truncation, so no extra > color maps are needed. > To allow storing into any depth will require subsequent addition > of two other > colormaps, as is the case with WarpBlt." > > | pixMask destShifted sourceShifted destPixVal rgbMask > sourcePixVal unAlpha result pixBlend shift blend maskShifted bitsPerColor > blendAG blendRB | > <inline: false> > + <returnTypeC: 'unsigned int'> > - <return: 'unsigned int'> > <var: #sourceWord type: 'unsigned int'> > <var: #destinationWord type: 'unsigned int'> > <var: #blendRB type: 'unsigned int'> > <var: #blendAG type: 'unsigned int'> > <var: #result type: 'unsigned int'> > <var: #sourceAlpha type: 'unsigned int'> > <var: #unAlpha type: 'unsigned int'> > + <var: #sourceShifted type: 'unsigned int'> > + <var: #destShifted type: 'unsigned int'> > + <var: #maskShifted type: 'unsigned int'> > + <var: #pixMask type: 'unsigned int'> > + <var: #rgbMask type: 'unsigned int'> > + <var: #pixBlend type: 'unsigned int'> > + <var: #blend type: 'unsigned int'> > destDepth < 16 ifTrue: [^ destinationWord "no-op"]. > unAlpha := 255 - sourceAlpha. > result := destinationWord. > destPPW = 1 ifTrue:["32bpp blends include alpha" > paintMode & (sourceWord = 0) "painting a transparent > pixel" ifFalse:[ > > blendRB := ((sourceWord bitAnd: 16rFF00FF) > * sourceAlpha) + > ((destinationWord bitAnd: > 16rFF00FF) * unAlpha) + 16rFF00FF. "blendRB red and blue" > > blendAG := ((sourceWord>> 8 bitAnd: > 16rFF00FF) * sourceAlpha) + > ((destinationWord>>8 > bitAnd: 16rFF00FF) * unAlpha) + 16rFF00FF. "blendRB alpha and green" > > blendRB := blendRB + (blendRB - 16r10001 > >> 8 bitAnd: 16rFF00FF) >> 8 bitAnd: 16rFF00FF. "divide by 255" > blendAG := blendAG + (blendAG - 16r10001 > >> 8 bitAnd: 16rFF00FF) >> 8 bitAnd: 16rFF00FF. > result := blendRB bitOr: blendAG<<8. > ]. > ] ifFalse:[ > pixMask := maskTable at: destDepth. > bitsPerColor := 5. > rgbMask := 16r1F. > maskShifted := destMask. > destShifted := destinationWord. > sourceShifted := sourceWord. > 1 to: destPPW do:[:j | > sourcePixVal := sourceShifted bitAnd: pixMask. > ((maskShifted bitAnd: pixMask) = 0 "no effect if > outside of dest rectangle" > or: [paintMode & (sourcePixVal = 0) "or > painting a transparent pixel"]) > ifFalse: > [destPixVal := destShifted bitAnd: pixMask. > pixBlend := 0. > 1 to: 3 do: > [:i | shift := (i-1)*bitsPerColor. > blend := (((sourcePixVal>>shift > bitAnd: rgbMask) * sourceAlpha) > + > ((destPixVal>>shift bitAnd: rgbMask) * unAlpha)) > + 254 // 255 > bitAnd: rgbMask. > pixBlend := pixBlend bitOr: > blend<<shift]. > result := (result bitAnd: (pixMask << > (j-1*16)) bitInvert32) > bitOr: > pixBlend << (j-1*16)]. > maskShifted := maskShifted >> destDepth. > sourceShifted := sourceShifted >> destDepth. > destShifted := destShifted >> destDepth]. > ]. > ^ result > ! > > Item was changed: > ----- Method: BitBltSimulation>>alphaBlendScaled:with: (in category > 'combination rules') ----- > alphaBlendScaled: sourceWord with: destinationWord > "Blend sourceWord with destinationWord using the alpha value from > sourceWord. > Alpha is encoded as 0 meaning 0.0, and 255 meaning 1.0. > In contrast to alphaBlend:with: the color produced is > > srcColor + (1-srcAlpha) * dstColor > > e.g., it is assumed that the source color is already scaled." > | unAlpha rb ag | > <inline: false> "Do NOT inline this into optimized loops" > + <returnTypeC: 'unsigned int'> > <var: #sourceWord type: 'unsigned int'> > <var: #destinationWord type: 'unsigned int'> > <var: #rb type: 'unsigned int'> > <var: #ag type: 'unsigned int'> > <var: #unAlpha type: 'unsigned int'> > unAlpha := 255 - (sourceWord >> 24). "High 8 bits of source pixel > is source opacity (ARGB format)" > rb := ((destinationWord bitAnd: 16rFF00FF) * unAlpha >> 8 bitAnd: > 16rFF00FF) + (sourceWord bitAnd: 16rFF00FF). "blend red and blue components" > ag := ((destinationWord>>8 bitAnd: 16rFF00FF) * unAlpha >> 8 > bitAnd: 16rFF00FF) + (sourceWord>>8 bitAnd: 16rFF00FF). "blend alpha and > green components" > rb := (rb bitAnd: 16rFF00FF) bitOr: (rb bitAnd: 16r1000100) * > 16rFF >> 8. "saturate red and blue components if there is a carry" > ag := (ag bitAnd: 16rFF00FF) << 8 bitOr: (ag bitAnd: 16r1000100) * > 16rFF. "saturate alpha and green components if there is a carry" > ^ag bitOr: rb "recompose"! > > Item was changed: > ----- Method: BitBltSimulation>>alphaPaintConst:with: (in category > 'combination rules') ----- > alphaPaintConst: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > > sourceWord = 0 ifTrue: [^ destinationWord "opt for > all-transparent source"]. > ^ self alphaBlendConst: sourceWord with: destinationWord > paintMode: true! > > Item was changed: > ----- Method: BitBltSimulation>>alphaSourceBlendBits16 (in category > 'inner loop') ----- > alphaSourceBlendBits16 > "This version assumes > combinationRule = 34 > sourcePixSize = 32 > destPixSize = 16 > sourceForm ~= destForm. > " > | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY > srcY dstY dstMask srcShift ditherBase ditherIndex ditherThreshold | > <inline: false> "This particular method should be optimized in > itself" > + <var: #sourceWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > + <var: #dstMask type: #'unsigned int'> > deltaY := bbH + 1. "So we can pre-decrement" > srcY := sy. > dstY := dy. > srcShift := (dx bitAnd: 1) * 16. > destMSB ifTrue:[srcShift := 16 - srcShift]. > mask1 := 16rFFFF << (16 - srcShift). > "This is the outer loop" > [(deltaY := deltaY - 1) ~= 0] whileTrue:[ > srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). > dstIndex := destBits + (dstY * destPitch) + (dx // 2 * 4). > ditherBase := (dstY bitAnd: 3) * 4. > ditherIndex := (sx bitAnd: 3) - 1. "For pre-increment" > deltaX := bbW + 1. "So we can pre-decrement" > dstMask := mask1. > dstMask = 16rFFFF ifTrue:[srcShift := 16] > ifFalse:[srcShift := 0]. > > "This is the inner loop" > [(deltaX := deltaX - 1) ~= 0] whileTrue:[ > ditherThreshold := ditherMatrix4x4 at: ditherBase > + (ditherIndex := ditherIndex + 1 bitAnd: 3). > sourceWord := self srcLongAt: srcIndex. > srcAlpha := sourceWord >> 24. > srcAlpha = 255 ifTrue:[ > "Dither from 32 to 16 bit" > sourceWord := self dither32To16: > sourceWord threshold: ditherThreshold. > sourceWord = 0 > ifTrue:[sourceWord := 1 << > srcShift] > ifFalse: [sourceWord := sourceWord > << srcShift]. > "Store masked value" > self dstLongAt: dstIndex put: sourceWord > mask: dstMask. > ] ifFalse:[ "srcAlpha ~= 255" > srcAlpha = 0 ifFalse:[ "0 < srcAlpha < 255" > "If we have to mix colors then > just copy a single word" > destWord := self dstLongAt: > dstIndex. > destWord := destWord bitAnd: > dstMask bitInvert32. > destWord := destWord >> srcShift. > "Expand from 16 to 32 bit by > adding zero bits" > destWord := (((destWord bitAnd: > 16r7C00) bitShift: 9) bitOr: > > ((destWord bitAnd: 16r3E0) bitShift: 6)) bitOr: > > (((destWord bitAnd: 16r1F) bitShift: 3) bitOr: > > 16rFF000000). > "Mix colors" > sourceWord := self > alphaBlendScaled: sourceWord with: destWord. > "And dither" > sourceWord := self dither32To16: > sourceWord threshold: ditherThreshold. > sourceWord = 0 > ifTrue:[sourceWord := 1 << > srcShift] > ifFalse:[sourceWord := > sourceWord << srcShift]. > "Store back" > self dstLongAt: dstIndex put: > sourceWord mask: dstMask. > ]. > ]. > srcIndex := srcIndex + 4. > destMSB > ifTrue:[srcShift = 0 ifTrue:[dstIndex := > dstIndex + 4]] > ifFalse:[srcShift = 0 ifFalse:[dstIndex := > dstIndex + 4]]. > srcShift := srcShift bitXor: 16. "Toggle between 0 > and 16" > dstMask := dstMask bitInvert32. "Mask other half > word" > ]. > srcY := srcY + 1. > dstY := dstY + 1. > ].! > > Item was changed: > ----- Method: BitBltSimulation>>alphaSourceBlendBits32 (in category > 'inner loop') ----- > alphaSourceBlendBits32 > "This version assumes > combinationRule = 34 > sourcePixSize = destPixSize = 32 > sourceForm ~= destForm. > Note: The inner loop has been optimized for dealing > with the special cases of srcAlpha = 0.0 and srcAlpha = 1.0 > " > | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY > srcY dstY | > <inline: false> "This particular method should be optimized in > itself" > + <var: #sourceWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > > "Give the compile a couple of hints" > > "The following should be declared as pointers so the compiler will > notice that they're used for accessing memory locations > (good to know on an Intel architecture) but then the increments > would be different between ST code and C code so must hope the > compiler notices what happens (MS Visual C does)" > > deltaY := bbH + 1. "So we can pre-decrement" > srcY := sy. > dstY := dy. > > "This is the outer loop" > [(deltaY := deltaY - 1) ~= 0] whileTrue:[ > srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). > dstIndex := destBits + (dstY * destPitch) + (dx * 4). > deltaX := bbW + 1. "So we can pre-decrement" > > "This is the inner loop" > [(deltaX := deltaX - 1) ~= 0] whileTrue:[ > sourceWord := self srcLongAt: srcIndex. > srcAlpha := sourceWord >> 24. > srcAlpha = 255 ifTrue:[ > self dstLongAt: dstIndex put: sourceWord. > srcIndex := srcIndex + 4. > dstIndex := dstIndex + 4. > "Now copy as many words as possible with > alpha = 255" > [(deltaX := deltaX - 1) ~= 0 and:[ > (sourceWord := self srcLongAt: > srcIndex) >> 24 = 255]] > whileTrue:[ > self dstLongAt: > dstIndex put: sourceWord. > srcIndex := > srcIndex + 4. > dstIndex := > dstIndex + 4. > ]. > "Adjust deltaX" > deltaX := deltaX + 1. > ] ifFalse:[ "srcAlpha ~= 255" > srcAlpha = 0 ifTrue:[ > srcIndex := srcIndex + 4. > dstIndex := dstIndex + 4. > "Now skip as many words as > possible," > [(deltaX := deltaX - 1) ~= 0 and:[ > (sourceWord := self > srcLongAt: srcIndex) >> 24 = 0]] > whileTrue:[ > srcIndex := > srcIndex + 4. > dstIndex := > dstIndex + 4. > ]. > "Adjust deltaX" > deltaX := deltaX + 1. > ] ifFalse:[ "0 < srcAlpha < 255" > "If we have to mix colors then > just copy a single word" > destWord := self dstLongAt: > dstIndex. > destWord := self alphaBlendScaled: > sourceWord with: destWord. > self dstLongAt: dstIndex put: > destWord. > srcIndex := srcIndex + 4. > dstIndex := dstIndex + 4. > ]. > ]. > ]. > srcY := srcY + 1. > dstY := dstY + 1. > ].! > > Item was changed: > ----- Method: BitBltSimulation>>alphaSourceBlendBits8 (in category > 'inner loop') ----- > alphaSourceBlendBits8 > "This version assumes > combinationRule = 34 > sourcePixSize = 32 > destPixSize = 8 > sourceForm ~= destForm. > Note: This is not real blending since we don't have the source > colors available. > " > | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY > srcY dstY dstMask srcShift adjust mappingTable mapperFlags | > <inline: false> > <var: #mappingTable type:'unsigned int *'> > + <var: #sourceWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > + <var: #dstMask type: #'unsigned int'> > mappingTable := self default8To32Table. > mapperFlags := cmFlags bitAnd: ColorMapNewStyle bitInvert32. > deltaY := bbH + 1. "So we can pre-decrement" > srcY := sy. > dstY := dy. > mask1 := ((dx bitAnd: 3) * 8). > destMSB ifTrue:[mask1 := 24 - mask1]. > mask2 := AllOnes bitXor:(16rFF << mask1). > (dx bitAnd: 1) = 0 > ifTrue:[adjust := 0] > ifFalse:[adjust := 16r1F1F1F1F]. > (dy bitAnd: 1) = 0 > ifTrue:[adjust := adjust bitXor: 16r1F1F1F1F]. > "This is the outer loop" > [(deltaY := deltaY - 1) ~= 0] whileTrue:[ > adjust := adjust bitXor: 16r1F1F1F1F. > srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). > dstIndex := destBits + (dstY * destPitch) + (dx // 4 * 4). > deltaX := bbW + 1. "So we can pre-decrement" > srcShift := mask1. > dstMask := mask2. > > "This is the inner loop" > [(deltaX := deltaX - 1) ~= 0] whileTrue:[ > sourceWord := ((self srcLongAt: srcIndex) bitAnd: > (adjust bitInvert32)) + adjust. > srcAlpha := sourceWord >> 24. > srcAlpha > 31 ifTrue:["Everything below 31 is > transparent" > srcAlpha < 224 ifTrue:["Everything above > 224 is opaque" > destWord := self dstLongAt: > dstIndex. > destWord := destWord bitAnd: > dstMask bitInvert32. > destWord := destWord >> srcShift. > destWord := mappingTable at: > destWord. > sourceWord := self > alphaBlendScaled: sourceWord with: destWord. > ]. > sourceWord := self mapPixel: sourceWord > flags: mapperFlags. > sourceWord := sourceWord << srcShift. > "Store back" > self dstLongAt: dstIndex put: sourceWord > mask: dstMask. > ]. > srcIndex := srcIndex + 4. > destMSB ifTrue:[ > srcShift = 0 > ifTrue:[dstIndex := dstIndex + 4. > srcShift := 24. > dstMask := > 16r00FFFFFF] > ifFalse:[srcShift := srcShift - 8. > dstMask := > (dstMask >> 8) bitOr: 16rFF000000]. > ] ifFalse:[ > srcShift = 24 > ifTrue:[dstIndex := dstIndex + 4. > srcShift := 0. > dstMask := > 16rFFFFFF00] > ifFalse:[srcShift := srcShift + 8. > dstMask := dstMask > << 8 bitOr: 255]. > ]. > adjust := adjust bitXor: 16r1F1F1F1F. > ]. > srcY := srcY + 1. > dstY := dstY + 1. > ].! > > Item was changed: > ----- Method: BitBltSimulation>>bitAnd:with: (in category 'combination > rules') ----- > bitAnd: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitAnd: destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>bitAndInvert:with: (in category > 'combination rules') ----- > bitAndInvert: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitAnd: destinationWord bitInvert32! > > Item was changed: > ----- Method: BitBltSimulation>>bitInvertAnd:with: (in category > 'combination rules') ----- > bitInvertAnd: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitInvert32 bitAnd: destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>bitInvertAndInvert:with: (in category > 'combination rules') ----- > bitInvertAndInvert: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitInvert32 bitAnd: destinationWord bitInvert32! > > Item was changed: > ----- Method: BitBltSimulation>>bitInvertDestination:with: (in category > 'combination rules') ----- > bitInvertDestination: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^destinationWord bitInvert32! > > Item was changed: > ----- Method: BitBltSimulation>>bitInvertOr:with: (in category > 'combination rules') ----- > bitInvertOr: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitInvert32 bitOr: destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>bitInvertOrInvert:with: (in category > 'combination rules') ----- > bitInvertOrInvert: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitInvert32 bitOr: destinationWord bitInvert32! > > Item was changed: > ----- Method: BitBltSimulation>>bitInvertSource:with: (in category > 'combination rules') ----- > bitInvertSource: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitInvert32! > > Item was changed: > ----- Method: BitBltSimulation>>bitInvertXor:with: (in category > 'combination rules') ----- > bitInvertXor: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitInvert32 bitXor: destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>bitOr:with: (in category 'combination > rules') ----- > bitOr: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitOr: destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>bitOrInvert:with: (in category > 'combination rules') ----- > bitOrInvert: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitOr: destinationWord bitInvert32! > > Item was changed: > ----- Method: BitBltSimulation>>bitXor:with: (in category 'combination > rules') ----- > bitXor: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord bitXor: destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>clearWord:with: (in category > 'combination rules') ----- > + clearWord: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > - clearWord: source with: destination > ^ 0! > > Item was changed: > ----- Method: BitBltSimulation>>copyLoop (in category 'inner loop') ----- > copyLoop > | prevWord thisWord skewWord halftoneWord mergeWord hInc y unskew > skewMask notSkewMask mergeFnwith destWord | > "This version of the inner loop assumes noSource = false." > <inline: false> > + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned > int, unsigned int)'> > + <var: #prevWord type: #'unsigned int'> > + <var: #thisWord type: #'unsigned int'> > + <var: #skewWord type: #'unsigned int'> > + <var: #halftoneWord type: #'unsigned int'> > + <var: #mergeWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > + <var: #skewMask type: #'unsigned int'> > + <var: #notSkewMask type: #'unsigned int'> > + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'unsigned int (*)(unsigned int, unsigned int)'. > - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> > - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'sqInt (*)(sqInt, sqInt)'. > mergeFnwith. "null ref for compiler" > > hInc := hDir*4. "Byte delta" > "degenerate skew fixed for Sparc. 10/20/96 ikp" > skew == -32 > ifTrue: [skew := unskew := skewMask := 0] > ifFalse: [skew < 0 > ifTrue: > [unskew := skew+32. > skewMask := AllOnes << (0-skew)] > ifFalse: > [skew = 0 > ifTrue: > [unskew := 0. > skewMask := AllOnes] > ifFalse: > [unskew := skew-32. > skewMask := AllOnes >> > skew]]]. > notSkewMask := skewMask bitInvert32. > noHalftone > ifTrue: [halftoneWord := AllOnes. halftoneHeight := 0] > ifFalse: [halftoneWord := self halftoneAt: 0]. > > y := dy. > 1 to: bbH do: "here is the vertical loop" > [ :i | > halftoneHeight > 1 ifTrue: "Otherwise, its always the > same" > [halftoneWord := self halftoneAt: y. > y := y + vDir]. > preload ifTrue: > ["load the 64-bit shifter" > prevWord := self srcLongAt: sourceIndex. > self incSrcIndex: hInc] > ifFalse: > [prevWord := 0]. > > "Note: the horizontal loop has been expanded into three parts for > speed:" > > "This first section requires masking of the > destination store..." > destMask := mask1. > thisWord := self srcLongAt: sourceIndex. "pick up > next word" > self incSrcIndex: hInc. > skewWord := ((prevWord bitAnd: notSkewMask) > bitShift: unskew) > bitOr: "32-bit > rotate" > ((thisWord bitAnd: > skewMask) bitShift: skew). > prevWord := thisWord. > destWord := self dstLongAt: destIndex. > mergeWord := self mergeFn: (skewWord bitAnd: > halftoneWord) with: destWord. > destWord := (destMask bitAnd: mergeWord) bitOr: > (destWord bitAnd: > destMask bitInvert32). > self dstLongAt: destIndex put: destWord. > self incDestIndex: hInc. > > "This central horizontal loop requires no store masking" > destMask := AllOnes. > combinationRule = 3 > ifTrue: [(skew = 0) & (halftoneWord = AllOnes) > ifTrue: > ["Very special inner loop for STORE mode with no skew -- > just move words" > hDir = -1 > ifTrue: ["Woeful patch: revert to older code for hDir = -1" > 2 to: nWords-1 do: > [ :word | > thisWord := self srcLongAt: > sourceIndex. > self incSrcIndex: hInc. > self dstLongAt: destIndex put: > thisWord. > self incDestIndex: hInc]] > ifFalse: [2 to: nWords-1 do: > [ :word | "Note loop starts with > prevWord loaded (due to preload)" > self dstLongAt: destIndex put: > prevWord. > self incDestIndex: hInc. > prevWord := self srcLongAt: > sourceIndex. > self incSrcIndex: hInc]]] > ifFalse: > ["Special inner loop for STORE mode -- no need to call > merge" > 2 to: nWords-1 do: > [ :word | > thisWord := self srcLongAt: sourceIndex. > self incSrcIndex: hInc. > skewWord := ((prevWord bitAnd: notSkewMask) > bitShift: unskew) > bitOr: "32-bit > rotate" > ((thisWord bitAnd: > skewMask) bitShift: skew). > prevWord := thisWord. > self dstLongAt: destIndex put: (skewWord bitAnd: > halftoneWord). > self incDestIndex: hInc]] > ] ifFalse: [2 to: nWords-1 do: "Normal inner loop does merge:" > [ :word | > thisWord := self srcLongAt: sourceIndex. "pick up > next word" > self incSrcIndex: hInc. > skewWord := ((prevWord bitAnd: notSkewMask) > bitShift: unskew) > bitOr: "32-bit > rotate" > ((thisWord bitAnd: > skewMask) bitShift: skew). > prevWord := thisWord. > mergeWord := self mergeFn: (skewWord bitAnd: > halftoneWord) > with: (self > dstLongAt: destIndex). > self dstLongAt: destIndex put: mergeWord. > self incDestIndex: hInc] > ]. > > "This last section, if used, requires masking of the > destination store..." > nWords > 1 ifTrue: > [destMask := mask2. > thisWord := self srcLongAt: sourceIndex. "pick up > next word" > self incSrcIndex: hInc. > skewWord := ((prevWord bitAnd: notSkewMask) > bitShift: unskew) > bitOr: "32-bit > rotate" > ((thisWord bitAnd: > skewMask) bitShift: skew). > destWord := self dstLongAt: destIndex. > mergeWord := self mergeFn: (skewWord bitAnd: > halftoneWord) with: destWord. > destWord := (destMask bitAnd: mergeWord) bitOr: > (destWord bitAnd: > destMask bitInvert32). > self dstLongAt: destIndex put: destWord. > self incDestIndex: hInc]. > > self incSrcIndex: sourceDelta. > self incDestIndex: destDelta]! > > Item was changed: > ----- Method: BitBltSimulation>>copyLoopNoSource (in category 'inner > loop') ----- > copyLoopNoSource > "Faster copyLoop when source not used. hDir and vDir are both > positive, and perload and skew are unused" > | halftoneWord mergeWord mergeFnwith destWord | > <inline: false> > + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned > int, unsigned int)'> > + <var: #halftoneWord type: #'unsigned int'> > + <var: #mergeWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'unsigned int (*)(unsigned int, unsigned int)'. > - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> > - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'sqInt (*)(sqInt, sqInt)'. > mergeFnwith. "null ref for compiler" > > 1 to: bbH do: "here is the vertical loop" > [ :i | > noHalftone > ifTrue: [halftoneWord := AllOnes] > ifFalse: [halftoneWord := self halftoneAt: dy+i-1]. > > "Note: the horizontal loop has been expanded into three parts for > speed:" > > "This first section requires masking of the > destination store..." > destMask := mask1. > destWord := self dstLongAt: destIndex. > mergeWord := self mergeFn: halftoneWord > with: destWord. > destWord := (destMask bitAnd: mergeWord) bitOr: > (destWord bitAnd: > destMask bitInvert32). > self dstLongAt: destIndex put: destWord. > self incDestIndex: 4. > > "This central horizontal loop requires no store masking" > destMask := AllOnes. > combinationRule = 3 ifTrue: ["Special inner loop > for STORE" > destWord := halftoneWord. > 2 to: nWords-1 do:[ :word | > self dstLongAt: destIndex put: > destWord. > self incDestIndex: 4]. > ] ifFalse:[ "Normal inner loop does merge" > 2 to: nWords-1 do:[ :word | "Normal inner > loop does merge" > destWord := self dstLongAt: > destIndex. > mergeWord := self mergeFn: > halftoneWord with: destWord. > self dstLongAt: destIndex put: > mergeWord. > self incDestIndex: 4]. > ]. > > "This last section, if used, requires masking of the > destination store..." > nWords > 1 ifTrue: > [destMask := mask2. > destWord := self dstLongAt: destIndex. > mergeWord := self mergeFn: halftoneWord with: > destWord. > destWord := (destMask bitAnd: mergeWord) bitOr: > (destWord bitAnd: > destMask bitInvert32). > self dstLongAt: destIndex put: destWord. > self incDestIndex: 4]. > > self incDestIndex: destDelta]! > > Item was changed: > ----- Method: BitBltSimulation>>copyLoopPixMap (in category 'inner > loop') ----- > copyLoopPixMap > "This version of the inner loop maps source pixels > to a destination form with different depth. Because it is already > unweildy, the loop is not unrolled as in the other versions. > Preload, skew and skewMask are all overlooked, since > pickSourcePixels > delivers its destination word already properly aligned. > Note that pickSourcePixels could be copied in-line at the top of > the horizontal loop, and some of its inits moved out of the loop." > "ar 12/7/1999: > The loop has been rewritten to use only one pickSourcePixels call. > The idea is that the call itself could be inlined. If we decide not > to inline pickSourcePixels we could optimize the loop instead." > | skewWord halftoneWord mergeWord scrStartBits nSourceIncs > startBits endBits sourcePixMask destPixMask mergeFnwith nPix srcShift > dstShift destWord words srcShiftInc dstShiftInc dstShiftLeft mapperFlags | > <inline: false> > + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned > int, unsigned int)'> > + <var: #skewWord type: #'unsigned int'> > + <var: #halftoneWord type: #'unsigned int'> > + <var: #mergeWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > + <var: #sourcePixMask type: #'unsigned int'> > + <var: #destPixMask type: #'unsigned int'> > + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'unsigned int (*)(unsigned int, unsigned int)'. > - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> > - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'sqInt (*)(sqInt, sqInt)'. > mergeFnwith. "null ref for compiler" > > "Additional inits peculiar to unequal source and dest pix size..." > sourcePPW := 32//sourceDepth. > sourcePixMask := maskTable at: sourceDepth. > destPixMask := maskTable at: destDepth. > mapperFlags := cmFlags bitAnd: ColorMapNewStyle bitInvert32. > sourceIndex := sourceBits + > (sy * sourcePitch) + ((sx // > sourcePPW) *4). > scrStartBits := sourcePPW - (sx bitAnd: sourcePPW-1). > bbW < scrStartBits > ifTrue: [nSourceIncs := 0] > ifFalse: [nSourceIncs := (bbW - scrStartBits)//sourcePPW + > 1]. > sourceDelta := sourcePitch - (nSourceIncs * 4). > > "Note following two items were already calculated in destmask > setup!!" > startBits := destPPW - (dx bitAnd: destPPW-1). > endBits := ((dx + bbW - 1) bitAnd: destPPW-1) + 1. > > bbW < startBits ifTrue:[startBits := bbW]. > > "Precomputed shifts for pickSourcePixels" > srcShift := ((sx bitAnd: sourcePPW - 1) * sourceDepth). > dstShift := ((dx bitAnd: destPPW - 1) * destDepth). > srcShiftInc := sourceDepth. > dstShiftInc := destDepth. > dstShiftLeft := 0. > sourceMSB ifTrue:[ > srcShift := 32 - sourceDepth - srcShift. > srcShiftInc := 0 - srcShiftInc]. > destMSB ifTrue:[ > dstShift := 32 - destDepth - dstShift. > dstShiftInc := 0 - dstShiftInc. > dstShiftLeft := 32 - destDepth]. > > 1 to: bbH do: "here is the vertical loop" > [ :i | > "*** is it possible at all that noHalftone == false? ***" > noHalftone > ifTrue:[halftoneWord := AllOnes] > ifFalse: [halftoneWord := self halftoneAt: dy+i-1]. > "setup first load" > srcBitShift := srcShift. > dstBitShift := dstShift. > destMask := mask1. > nPix := startBits. > "Here is the horizontal loop..." > words := nWords. > ["pick up the word" > skewWord := self pickSourcePixels: nPix flags: > mapperFlags > srcMask: > sourcePixMask destMask: destPixMask > > srcShiftInc: srcShiftInc dstShiftInc: dstShiftInc. > "align next word to leftmost pixel" > dstBitShift := dstShiftLeft. > > destMask = AllOnes ifTrue:["avoid > read-modify-write" > mergeWord := self mergeFn: (skewWord > bitAnd: halftoneWord) > with: > (self dstLongAt: destIndex). > self dstLongAt: destIndex put: (destMask > bitAnd: mergeWord). > ] ifFalse:[ "General version using dest masking" > destWord := self dstLongAt: destIndex. > mergeWord := self mergeFn: (skewWord > bitAnd: halftoneWord) > with: > (destWord bitAnd: destMask). > destWord := (destMask bitAnd: mergeWord) > bitOr: > (destWord > bitAnd: destMask bitInvert32). > self dstLongAt: destIndex put: destWord. > ]. > self incDestIndex: 4. > words = 2 "e.g., is the next word the last word?" > ifTrue:["set mask for last word in this > row" > destMask := mask2. > nPix := endBits] > ifFalse:["use fullword mask for inner loop" > destMask := AllOnes. > nPix := destPPW]. > (words := words - 1) = 0] whileFalse. > "--- end of inner loop ---" > self incSrcIndex: sourceDelta. > self incDestIndex: destDelta] > ! > > Item was changed: > ----- Method: BitBltSimulation>>destinationWord:with: (in category > 'combination rules') ----- > destinationWord: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>dither32To16:threshold: (in category > 'pixel mapping') ----- > dither32To16: srcWord threshold: ditherValue > "Dither the given 32bit word to 16 bit. Ignore alpha." > | addThreshold | > <inline: true> "You bet" > + <returnTypeC: 'unsigned int'> > + <var: #srcWord type: 'unsigned int'> > addThreshold := ditherValue bitShift: 8. > ^((dither8Lookup at: (addThreshold+((srcWord bitShift: -16) > bitAnd: 255))) bitShift: 10) + > ((dither8Lookup at: (addThreshold+((srcWord bitShift: -8) > bitAnd: 255))) bitShift: 5) + > (dither8Lookup at: (addThreshold+(srcWord bitAnd: 255))). > ! > > Item was changed: > ----- Method: BitBltSimulation>>expensiveDither32To16:threshold: (in > category 'pixel mapping') ----- > expensiveDither32To16: srcWord threshold: ditherValue > "Dither the given 32bit word to 16 bit. Ignore alpha." > | pv threshold value out | > <inline: true> "You bet" > + <returnTypeC: 'unsigned int'> > + <var: #srcWord type: 'unsigned int'> > pv := srcWord bitAnd: 255. > threshold := ditherThresholds16 at: (pv bitAnd: 7). > value := ditherValues16 at: (pv bitShift: -3). > ditherValue < threshold > ifTrue:[out := value + 1] > ifFalse:[out := value]. > pv := (srcWord bitShift: -8) bitAnd: 255. > threshold := ditherThresholds16 at: (pv bitAnd: 7). > value := ditherValues16 at: (pv bitShift: -3). > ditherValue < threshold > ifTrue:[out := out bitOr: (value+1 bitShift:5)] > ifFalse:[out := out bitOr: (value bitShift: 5)]. > pv := (srcWord bitShift: -16) bitAnd: 255. > threshold := ditherThresholds16 at: (pv bitAnd: 7). > value := ditherValues16 at: (pv bitShift: -3). > ditherValue < threshold > ifTrue:[out := out bitOr: (value+1 bitShift:10)] > ifFalse:[out := out bitOr: (value bitShift: 10)]. > ^out! > > Item was changed: > ----- Method: BitBltSimulation>>fixAlpha:with: (in category 'combination > rules') ----- > fixAlpha: sourceWord with: destinationWord > "For any non-zero pixel value in destinationWord with zero alpha > channel take the alpha from sourceWord and fill it in. Intended for fixing > alpha channels left at zero during 16->32 bpp conversions." > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > destDepth = 32 ifFalse:[^destinationWord]. "no-op for non 32bpp" > destinationWord = 0 ifTrue:[^0]. > (destinationWord bitAnd: 16rFF000000) = 0 > ifFalse:[^destinationWord]. > ^destinationWord bitOr: (sourceWord bitAnd: 16rFF000000) > ! > > Item was changed: > ----- Method: BitBltSimulation>>merge:with: (in category 'combination > rules') ----- > merge: sourceWord with: destinationWord > | mergeFnwith | > "Sender warpLoop is too big to include this in-line" > + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned > int, unsigned int)'> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'unsigned int (*)(unsigned int, unsigned int)'. > - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> > - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'sqInt (*)(sqInt, sqInt)'. > mergeFnwith. "null ref for compiler" > > ^ self mergeFn: sourceWord with: destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>partitionedAND:to:nBits:nPartitions: (in > category 'combination rules') ----- > partitionedAND: word1 to: word2 nBits: nBits nPartitions: nParts > "AND word1 to word2 as nParts partitions of nBits each. > Any field of word1 not all-ones is treated as all-zeroes. > Used for erasing, eg, brush shapes prior to ORing in a color" > | mask result | > + <returnTypeC: 'unsigned int'> > + <var: #word1 type: 'unsigned int'> > + <var: #word2 type: 'unsigned int'> > + <var: #result type: 'unsigned int'> > mask := maskTable at: nBits. "partition mask starts at the right" > result := 0. > 1 to: nParts do: > [:i | > (word1 bitAnd: mask) = mask > ifTrue: [result := result bitOr: (word2 bitAnd: > mask)]. > mask := mask << nBits "slide left to next partition"]. > ^ result > ! > > Item was changed: > ----- Method: > BitBltSimulation>>partitionedAdd:to:nBits:componentMask:carryOverflowMask: > (in category 'combination rules') ----- > partitionedAdd: word1 to: word2 nBits: nBits componentMask: > componentMask carryOverflowMask: carryOverflowMask > "Add word1 to word2 as nParts partitions of nBits each. > This is useful for packed pixels, or packed colors" > | carryOverflow sum w1 w2 | > "Use unsigned int everywhere because it has a well known > arithmetic model without undefined behavior w.r.t. overflow and shifts" > + <returnTypeC: 'unsigned int'> > <var: #word1 type: 'unsigned int'> > <var: #word2 type: 'unsigned int'> > <var: #w1 type: 'unsigned int'> > <var: #w2 type: 'unsigned int'> > <var: #componentMask type: 'unsigned int'> > <var: #carryOverflowMask type: 'unsigned int'> > <var: #carryOverflow type: 'unsigned int'> > <var: #sum type: 'unsigned int'> > w1 := word1 bitAnd: carryOverflowMask. "mask to remove high bit of > each component" > w2 := word2 bitAnd: carryOverflowMask. > sum := (word1 bitXor: w1)+(word2 bitXor: w2). "sum without high > bit to avoid overflowing over next component" > carryOverflow := (w1 bitAnd: w2) bitOr: ((w1 bitOr: w2) bitAnd: > sum). "detect overflow condition for saturating" > ^((sum bitXor: w1)bitXor:w2) "sum high bit without overflow" > bitOr: carryOverflow>>(nBits-1) * componentMask "saturate > in case of overflow"! > > Item was changed: > ----- Method: BitBltSimulation>>partitionedMax:with:nBits:nPartitions: > (in category 'combination rules') ----- > partitionedMax: word1 with: word2 nBits: nBits nPartitions: nParts > "Max word1 to word2 as nParts partitions of nBits each" > | mask result | > "In C, most arithmetic operations answer the same bit pattern > regardless of the operands being signed or unsigned ints > (this is due to the way 2's complement numbers work). However, > comparisions might fail. Add the proper declaration of > words as unsigned int in those cases where comparisions are done > (jmv)" > + <returnTypeC: 'unsigned int'> > <var: #word1 type: 'unsigned int'> > <var: #word2 type: 'unsigned int'> > <var: #mask type: 'unsigned int'> > <var: #result type: 'unsigned int'> > mask := maskTable at: nBits. "partition mask starts at the right" > result := 0. > 1 to: nParts do: > [:i | > result := result bitOr: ((word2 bitAnd: mask) max: (word1 > bitAnd: mask)). > mask := mask << nBits "slide left to next partition"]. > ^ result > ! > > Item was changed: > ----- Method: BitBltSimulation>>partitionedMin:with:nBits:nPartitions: > (in category 'combination rules') ----- > partitionedMin: word1 with: word2 nBits: nBits nPartitions: nParts > "Min word1 to word2 as nParts partitions of nBits each" > | mask result | > "In C, most arithmetic operations answer the same bit pattern > regardless of the operands being signed or unsigned ints > (this is due to the way 2's complement numbers work). However, > comparisions might fail. Add the proper declaration of > words as unsigned int in those cases where comparisions are done > (jmv)" > + <returnTypeC: 'unsigned int'> > <var: #word1 type: 'unsigned int'> > <var: #word2 type: 'unsigned int'> > <var: #mask type: 'unsigned int'> > <var: #result type: 'unsigned int'> > mask := maskTable at: nBits. "partition mask starts at the right" > result := 0. > 1 to: nParts do: > [:i | > result := result bitOr: ((word2 bitAnd: mask) min: (word1 > bitAnd: mask)). > mask := mask << nBits "slide left to next partition"]. > ^ result > ! > > Item was changed: > ----- Method: BitBltSimulation>>partitionedMul:with:nBits:nPartitions: > (in category 'combination rules') ----- > partitionedMul: word1 with: word2 nBits: nBits nPartitions: nParts > "Multiply word1 with word2 as nParts partitions of nBits each. > This is useful for packed pixels, or packed colors. > Bug in loop version when non-white background" > > | sMask product result dMask | > "In C, integer multiplication might answer a wrong value if the > unsigned values are declared as signed. > This problem does not affect this method, because the most > significant bit (i.e. the sign bit) will > always be zero (jmv)" > + <returnTypeC: 'unsigned int'> > + <var: #word1 type: 'unsigned int'> > + <var: #word2 type: 'unsigned int'> > + <var: #sMask type: 'unsigned int'> > + <var: #dMask type: 'unsigned int'> > + <var: #result type: 'unsigned int'> > + <var: #product type: 'unsigned int'> > sMask := maskTable at: nBits. "partition mask starts at the right" > dMask := sMask << nBits. > result := (((word1 bitAnd: sMask)+1) * ((word2 bitAnd: sMask)+1) - > 1 > bitAnd: dMask) >> nBits. "optimized > first step" > nParts = 1 > ifTrue: [ ^result ]. > product := (((word1>>nBits bitAnd: sMask)+1) * ((word2>>nBits > bitAnd: sMask)+1) - 1 bitAnd: dMask). > result := result bitOr: product. > nParts = 2 > ifTrue: [ ^result ]. > product := (((word1>>(2*nBits) bitAnd: sMask)+1) * > ((word2>>(2*nBits) bitAnd: sMask)+1) - 1 bitAnd: dMask). > result := result bitOr: product << nBits. > nParts = 3 > ifTrue: [ ^result ]. > product := (((word1>>(3*nBits) bitAnd: sMask)+1) * > ((word2>>(3*nBits) bitAnd: sMask)+1) - 1 bitAnd: dMask). > result := result bitOr: product << (2*nBits). > ^ result > > " | sMask product result dMask | > sMask := maskTable at: nBits. 'partition mask starts at the right' > dMask := sMask << nBits. > result := (((word1 bitAnd: sMask)+1) * ((word2 bitAnd: sMask)+1) - > 1 > bitAnd: dMask) >> nBits. 'optimized > first step' > nBits to: nBits * (nParts-1) by: nBits do: [:ofs | > product := (((word1>>ofs bitAnd: sMask)+1) * ((word2>>ofs > bitAnd: sMask)+1) - 1 bitAnd: dMask). > result := result bitOr: (product bitAnd: dMask) << > (ofs-nBits)]. > ^ result"! > > Item was changed: > ----- Method: > BitBltSimulation>>partitionedRgbComponentAlpha:dest:nBits:nPartitions: (in > category 'combination rules') ----- > partitionedRgbComponentAlpha: sourceWord dest: destWord nBits: nBits > nPartitions: nParts > | mask result p1 p2 v | > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destWord type: 'unsigned int'> > + <var: #p1 type: 'unsigned int'> > + <var: #p2 type: 'unsigned int'> > + <var: #mask type: 'unsigned int'> > + <var: #result type: 'unsigned int'> > mask := maskTable at: nBits. "partition mask starts at the right" > result := 0. > 1 to: nParts do: > [:i | > p1 := (sourceWord bitAnd: mask) >> ((i - 1)*nBits). > p2 := (destWord bitAnd: mask) >> ((i - 1)*nBits). > nBits = 32 > ifFalse:[ > nBits = 16 > ifTrue:[ > p1 := (self rgbMap16To32: > p1) bitOr: 16rFF000000. > p2 := (self rgbMap16To32: > p2) bitOr: 16rFF000000] > ifFalse:[ > p1 := (self rgbMap: p1 > from: nBits to: 32) bitOr: 16rFF000000. > p2 := (self rgbMap: p2 > from: nBits to: 32) bitOr: 16rFF000000.]]. > v := self rgbComponentAlpha32: p1 with: p2. > nBits = 32 > ifFalse:[ > v := self rgbMap: v from: 32 to: nBits]. > result := result bitOr: (v << ((i - 1)*nBits)). > mask := mask << nBits "slide left to next partition"]. > ^ result > ! > > Item was changed: > ----- Method: BitBltSimulation>>partitionedSub:from:nBits:nPartitions: > (in category 'combination rules') ----- > partitionedSub: word1 from: word2 nBits: nBits nPartitions: nParts > "Subtract word1 from word2 as nParts partitions of nBits each. > This is useful for packed pixels, or packed colors" > | mask result p1 p2 | > "In C, most arithmetic operations answer the same bit pattern > regardless of the operands being signed or unsigned ints > (this is due to the way 2's complement numbers work). However, > comparisions might fail. Add the proper declaration of > words as unsigned int in those cases where comparisions are done > (jmv)" > + <returnTypeC: 'unsigned int'> > <var: #word1 type: 'unsigned int'> > <var: #word2 type: 'unsigned int'> > <var: #p1 type: 'unsigned int'> > <var: #p2 type: 'unsigned int'> > <var: #mask type: 'unsigned int'> > <var: #result type: 'unsigned int'> > mask := maskTable at: nBits. "partition mask starts at the right" > result := 0. > 1 to: nParts do: > [:i | > p1 := word1 bitAnd: mask. > p2 := word2 bitAnd: mask. > p1 < p2 "result is really abs value of thedifference" > ifTrue: [result := result bitOr: p2 - p1] > ifFalse: [result := result bitOr: p1 - p2]. > mask := mask << nBits "slide left to next partition"]. > ^ result > ! > > Item was changed: > ----- Method: > BitBltSimulation>>pickSourcePixels:flags:srcMask:destMask:srcShiftInc:dstShiftInc: > (in category 'combination rules') ----- > pickSourcePixels: nPixels flags: mapperFlags srcMask: srcMask destMask: > dstMask srcShiftInc: srcShiftInc dstShiftInc: dstShiftInc > "Pick nPix pixels starting at srcBitIndex from the source, map by > the > color map, and justify them according to dstBitIndex in the > resulting destWord." > | sourceWord destWord sourcePix destPix srcShift dstShift nPix | > <inline: true> "oh please" > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > sourceWord := self srcLongAt: sourceIndex. > destWord := 0. > srcShift := srcBitShift. "Hint: Keep in register" > dstShift := dstBitShift. "Hint: Keep in register" > nPix := nPixels. "always > 0 so we can use do { } while(--nPix);" > (mapperFlags = (ColorMapPresent bitOr: ColorMapIndexedPart)) > ifTrue:[ > "a little optimization for (pretty crucial) blits using > indexed lookups only" > [ "grab, colormap and mix in pixel" > sourcePix := sourceWord >> srcShift bitAnd: > srcMask. > destPix := cmLookupTable at: (sourcePix bitAnd: > cmMask). > destWord := destWord bitOr: (destPix bitAnd: > dstMask) << dstShift. > "adjust dest pix index" > dstShift := dstShift + dstShiftInc. > "adjust source pix index" > ((srcShift := srcShift + srcShiftInc) bitAnd: > 16rFFFFFFE0) = 0 ifFalse:[ > sourceMSB ifTrue:[srcShift := srcShift + > 32] ifFalse:[srcShift := srcShift - 32]. > sourceWord := self srcLongAt: (self > incSrcIndex: 4)]. > (nPix := nPix - 1) = 0] whileFalse. > ] ifFalse:[ > [ "grab, colormap and mix in pixel" > sourcePix := sourceWord >> srcShift bitAnd: > srcMask. > destPix := self mapPixel: sourcePix flags: > mapperFlags. > destWord := destWord bitOr: (destPix bitAnd: > dstMask) << dstShift. > "adjust dest pix index" > dstShift := dstShift + dstShiftInc. > "adjust source pix index" > ((srcShift := srcShift + srcShiftInc) bitAnd: > 16rFFFFFFE0) = 0 ifFalse:[ > sourceMSB ifTrue:[srcShift := srcShift + > 32] ifFalse:[srcShift := srcShift - 32]. > sourceWord := self srcLongAt: (self > incSrcIndex: 4)]. > (nPix := nPix - 1) = 0] whileFalse. > ]. > srcBitShift := srcShift. "Store back" > ^destWord > ! > > Item was changed: > ----- Method: BitBltSimulation>>pickWarpPixelAtX:y: (in category 'pixel > mapping') ----- > pickWarpPixelAtX: xx y: yy > "Pick a single pixel from the source for WarpBlt. > Note: This method is crucial for WarpBlt speed w/o smoothing > and still relatively important when smoothing is used." > | x y srcIndex sourceWord sourcePix | > <inline: true> "*please*" > + <returnTypeC: #'unsigned int'> > + <var: #sourceWord type: #'unsigned int'> > > "note: it would be much faster if we could just > avoid these stupid tests for being inside sourceForm." > (xx < 0 or:[yy < 0 or:[ > (x := xx >> BinaryPoint) >= sourceWidth or:[ > (y := yy >> BinaryPoint) >= sourceHeight]]]) > ifTrue:[^0]. "out of bounds" > > "Fetch source word. > Note: We should really update srcIndex with sx and sy so that > we don't have to do the computation below. We might even be > able to simplify the out of bounds test from above." > srcIndex := sourceBits + (y * sourcePitch) + (x >> warpAlignShift > * 4). > sourceWord := self srcLongAt: srcIndex. > > "Extract pixel from word" > srcBitShift := warpBitShiftTable at: (x bitAnd: warpAlignMask). > sourcePix := sourceWord >> srcBitShift bitAnd: warpSrcMask. > ^sourcePix! > > Item was changed: > ----- Method: BitBltSimulation>>pixClear:with: (in category 'combination > rules') ----- > pixClear: sourceWord with: destinationWord > "Clear all pixels in destinationWord for which the pixels of > sourceWord have the same values. Used to clear areas of some constant color > to zero." > | mask result nBits pv | > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > + <var: #mask type: 'unsigned int'> > + <var: #pv type: 'unsigned int'> > + <var: #result type: 'unsigned int'> > destDepth = 32 ifTrue:[ > sourceWord = destinationWord ifTrue:[^0] > ifFalse:[^destinationWord]. > ]. > nBits := destDepth. > mask := maskTable at: nBits. "partition mask starts at the right" > result := 0. > 1 to: destPPW do:[:i | > pv := destinationWord bitAnd: mask. > (sourceWord bitAnd: mask) = pv ifTrue:[pv := 0]. > result := result bitOr: pv. > mask := mask << nBits "slide left to next partition"]. > ^ result! > > Item was changed: > ----- Method: BitBltSimulation>>pixMask:with: (in category 'combination > rules') ----- > pixMask: sourceWord with: destinationWord > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^ self partitionedAND: sourceWord bitInvert32 to: destinationWord > nBits: destDepth nPartitions: > destPPW! > > Item was changed: > ----- Method: BitBltSimulation>>pixPaint:with: (in category 'combination > rules') ----- > pixPaint: sourceWord with: destinationWord > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > sourceWord = 0 ifTrue: [^ destinationWord]. > ^ sourceWord bitOr: > (self partitionedAND: sourceWord bitInvert32 to: > destinationWord > nBits: destDepth > nPartitions: destPPW)! > > Item was changed: > ----- Method: BitBltSimulation>>pixSwap:with: (in category 'combination > rules') ----- > pixSwap: sourceWord with: destWord > "Swap the pixels in destWord" > | result shift lowMask highMask | > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destWord type: 'unsigned int'> > + <var: #lowMask type: 'unsigned int'> > + <var: #highMask type: 'unsigned int'> > + <var: #result type: 'unsigned int'> > destPPW = 1 ifTrue:[^destWord]. "a single pixel per word" > result := 0. > lowMask := (1 << destDepth) - 1. "mask low pixel" > highMask := lowMask << (destPPW-1 * destDepth). "mask high pixel" > shift := 32 - destDepth. > result := result bitOr: ( > (destWord bitAnd: lowMask) << shift bitOr: > (destWord bitAnd: highMask) >> > shift). > destPPW <= 2 ifTrue:[^result]. > 2 to: destPPW // 2 do:[:i| > lowMask := lowMask << destDepth. > highMask := highMask >> destDepth. > shift := shift - (destDepth * 2). > result := result bitOr: ( > (destWord bitAnd: lowMask) << > shift bitOr: > (destWord bitAnd: > highMask) >> shift)]. > ^result! > > Item was changed: > ----- Method: BitBltSimulation>>rgbAdd:with: (in category 'combination > rules') ----- > rgbAdd: sourceWord with: destinationWord > <inline: false> > + <returnTypeC: 'unsigned int'> > <var: #sourceWord type: 'unsigned int'> > <var: #destinationWord type: 'unsigned int'> > <var: #carryOverflowMask type: 'unsigned int'> > <var: #componentMask type: 'unsigned int'> > | componentMask carryOverflowMask | > destDepth < 16 ifTrue: > ["Add each pixel separately" > componentMask := 1<<destDepth-1. > carryOverflowMask := > 16rFFFFFFFF//componentMask<<(destDepth-1). > ^ self partitionedAdd: sourceWord to: destinationWord > nBits: destDepth > componentMask: componentMask carryOverflowMask: carryOverflowMask]. > destDepth = 16 ifTrue: > ["Add RGB components of each pixel separately" > componentMask := 16r1F. > carryOverflowMask := 16r42104210. > ^ (self partitionedAdd: (sourceWord bitAnd: 16r7FFF7FFF) > to: (destinationWord bitAnd: 16r7FFF7FFF) "make sure that the unused bit is > at 0" > nBits: 5 componentMask: > componentMask carryOverflowMask: carryOverflowMask)] > ifFalse: > ["Add RGBA components of the pixel separately" > componentMask := 16rFF. > carryOverflowMask := 16r80808080. > ^ self partitionedAdd: sourceWord to: destinationWord > nBits: 8 componentMask: > componentMask carryOverflowMask: carryOverflowMask]! > > Item was changed: > ----- Method: BitBltSimulation>>rgbComponentAlpha16 (in category > 'combination rules') ----- > rgbComponentAlpha16 > "This version assumes > combinationRule = 41 > sourcePixSize = 32 > destPixSize = 16 > sourceForm ~= destForm. > " > <inline: false> "This particular method should be optimized in > itself" > + <var: #sourceWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > + <var: #dstMask type: #'unsigned int'> > > | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY > srcY dstY dstMask srcShift ditherBase ditherIndex ditherThreshold | > > deltaY := bbH + 1. "So we can pre-decrement" > srcY := sy. > dstY := dy. > srcShift := (dx bitAnd: 1) * 16. > destMSB ifTrue:[srcShift := 16 - srcShift]. > mask1 := 16rFFFF << (16 - srcShift). > "This is the outer loop" > [(deltaY := deltaY - 1) ~= 0] whileTrue:[ > srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). > dstIndex := destBits + (dstY * destPitch) + (dx // 2 * 4). > ditherBase := (dstY bitAnd: 3) * 4. > ditherIndex := (sx bitAnd: 3) - 1. "For pre-increment" > deltaX := bbW + 1. "So we can pre-decrement" > dstMask := mask1. > dstMask = 16rFFFF ifTrue:[srcShift := 16] > ifFalse:[srcShift := 0]. > > "This is the inner loop" > [(deltaX := deltaX - 1) ~= 0] whileTrue:[ > ditherThreshold := ditherMatrix4x4 at: ditherBase > + (ditherIndex := ditherIndex + 1 bitAnd: 3). > sourceWord := self srcLongAt: srcIndex. > srcAlpha := sourceWord bitAnd: 16rFFFFFF. > srcAlpha = 0 ifFalse:[ "0 < srcAlpha" > "If we have to mix colors then > just copy a single word" > destWord := self dstLongAt: > dstIndex. > destWord := destWord bitAnd: > dstMask bitInvert32. > destWord := destWord >> srcShift. > "Expand from 16 to 32 bit by > adding zero bits" > destWord := (((destWord bitAnd: > 16r7C00) bitShift: 9) bitOr: > > ((destWord bitAnd: 16r3E0) bitShift: 6)) bitOr: > > (((destWord bitAnd: 16r1F) bitShift: 3) bitOr: > > 16rFF000000). > "Mix colors" > sourceWord := self > rgbComponentAlpha32: sourceWord with: destWord. > "And dither" > sourceWord := self dither32To16: > sourceWord threshold: ditherThreshold. > sourceWord = 0 > ifTrue:[sourceWord := 1 << > srcShift] > ifFalse:[sourceWord := > sourceWord << srcShift]. > "Store back" > self dstLongAt: dstIndex put: > sourceWord mask: dstMask. > ]. > srcIndex := srcIndex + 4. > destMSB > ifTrue:[srcShift = 0 ifTrue:[dstIndex := > dstIndex + 4]] > ifFalse:[srcShift = 0 ifFalse:[dstIndex := > dstIndex + 4]]. > srcShift := srcShift bitXor: 16. "Toggle between 0 > and 16" > dstMask := dstMask bitInvert32. "Mask other half > word" > ]. > srcY := srcY + 1. > dstY := dstY + 1. > ]. > ! > > Item was changed: > ----- Method: BitBltSimulation>>rgbComponentAlpha32 (in category > 'combination rules') ----- > rgbComponentAlpha32 > "This version assumes > combinationRule = 41 > sourcePixSize = destPixSize = 32 > sourceForm ~= destForm. > Note: The inner loop has been optimized for dealing > with the special case of aR = aG = aB = 0 > " > | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY > srcY dstY | > > <inline: false> "This particular method should be optimized in > itself" > > "Give the compile a couple of hints" > - <var: #sourceWord type: 'register long'> > <var: #deltaX type: 'register long'> > + <var: #sourceWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > > "The following should be declared as pointers so the compiler will > notice that they're used for accessing memory locations > (good to know on an Intel architecture) but then the increments > would be different between ST code and C code so must hope the > compiler notices what happens (MS Visual C does)" > <var: #srcIndex type: 'register long'> > <var: #dstIndex type: 'register long'> > > deltaY := bbH + 1. "So we can pre-decrement" > srcY := sy. > dstY := dy. > > "This is the outer loop" > [(deltaY := deltaY - 1) ~= 0] whileTrue:[ > srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). > dstIndex := destBits + (dstY * destPitch) + (dx * 4). > deltaX := bbW + 1. "So we can pre-decrement" > > "This is the inner loop" > [(deltaX := deltaX - 1) ~= 0] whileTrue:[ > sourceWord := self srcLongAt: srcIndex. > srcAlpha := sourceWord bitAnd:16rFFFFFF. > srcAlpha = 0 ifTrue:[ > srcIndex := srcIndex + 4. > dstIndex := dstIndex + 4. > "Now skip as many words as > possible," > [(deltaX := deltaX - 1) ~= 0 and:[ > ((sourceWord := self > srcLongAt: srcIndex) bitAnd:16rFFFFFF) = 0]] > whileTrue:[ > srcIndex := > srcIndex + 4. > dstIndex := > dstIndex + 4. > ]. > "Adjust deltaX" > deltaX := deltaX + 1. > ] ifFalse:[ "0 < srcAlpha" > "If we have to mix colors then > just copy a single word" > destWord := self dstLongAt: > dstIndex. > destWord := self > rgbComponentAlpha32: sourceWord with: destWord. > self dstLongAt: dstIndex put: > destWord. > srcIndex := srcIndex + 4. > dstIndex := dstIndex + 4. > ]. > ]. > srcY := srcY + 1. > dstY := dstY + 1. > ].! > > Item was changed: > ----- Method: BitBltSimulation>>rgbComponentAlpha8 (in category > 'combination rules') ----- > rgbComponentAlpha8 > "This version assumes > combinationRule = 41 > sourcePixSize = 32 > destPixSize = 8 > sourceForm ~= destForm. > Note: This is not real blending since we don't have the source > colors available. > " > > | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY > srcY dstY dstMask srcShift adjust mappingTable mapperFlags | > > <inline: false> "This particular method should be optimized in > itself" > <var: #mappingTable declareC:'unsigned int *mappingTable'> > + <var: #sourceWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > + <var: #dstMask type: #'unsigned int'> > > mappingTable := self default8To32Table. > mapperFlags := cmFlags bitAnd: ColorMapNewStyle bitInvert32. > deltaY := bbH + 1. "So we can pre-decrement" > srcY := sy. > dstY := dy. > mask1 := ((dx bitAnd: 3) * 8). > destMSB ifTrue:[mask1 := 24 - mask1]. > mask2 := AllOnes bitXor:(16rFF << mask1). > (dx bitAnd: 1) = 0 > ifTrue:[adjust := 0] > ifFalse:[adjust := 16r1F1F1F1F]. > (dy bitAnd: 1) = 0 > ifTrue:[adjust := adjust bitXor: 16r1F1F1F1F]. > "This is the outer loop" > [(deltaY := deltaY - 1) ~= 0] whileTrue:[ > adjust := adjust bitXor: 16r1F1F1F1F. > srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). > dstIndex := destBits + (dstY * destPitch) + (dx // 4 * 4). > deltaX := bbW + 1. "So we can pre-decrement" > srcShift := mask1. > dstMask := mask2. > > "This is the inner loop" > [(deltaX := deltaX - 1) ~= 0] whileTrue:[ > sourceWord := ((self srcLongAt: srcIndex) bitAnd: > (adjust bitInvert32)) + adjust. > srcAlpha := sourceWord bitAnd: 16rFFFFFF. > "set srcAlpha to the average of the 3 separate > aR,Ag,AB values" > srcAlpha := ((srcAlpha >> 16) + (srcAlpha >> 8 > bitAnd: 16rFF) + (srcAlpha bitAnd: 16rFF)) // 3. > srcAlpha > 31 ifTrue:["Everything below 31 is > transparent" > srcAlpha > 224 > ifTrue: ["treat everything above > 224 as opaque" > sourceWord := 16rFFFFFFFF]. > destWord := self dstLongAt: dstIndex. > destWord := destWord bitAnd: dstMask > bitInvert32. > destWord := destWord >> srcShift. > destWord := mappingTable at: destWord. > sourceWord := self rgbComponentAlpha32: > sourceWord with: destWord. > sourceWord := self mapPixel: sourceWord > flags: mapperFlags. > sourceWord := sourceWord << srcShift. > "Store back" > self dstLongAt: dstIndex put: sourceWord > mask: dstMask. > ]. > srcIndex := srcIndex + 4. > destMSB ifTrue:[ > srcShift = 0 > ifTrue:[dstIndex := dstIndex + 4. > srcShift := 24. > dstMask := > 16r00FFFFFF] > ifFalse:[srcShift := srcShift - 8. > dstMask := > (dstMask >> 8) bitOr: 16rFF000000]. > ] ifFalse:[ > srcShift = 32 > ifTrue:[dstIndex := dstIndex + 4. > srcShift := 0. > dstMask := > 16rFFFFFF00] > ifFalse:[srcShift := srcShift + 8. > dstMask := dstMask > << 8 bitOr: 255]. > ]. > adjust := adjust bitXor: 16r1F1F1F1F. > ]. > srcY := srcY + 1. > dstY := dstY + 1. > ]. > ! > > Item was changed: > ----- Method: BitBltSimulation>>rgbDiff:with: (in category 'combination > rules') ----- > rgbDiff: sourceWord with: destinationWord > "Subract the pixels in the source and destination, color by color, > and return the sum of the absolute value of all the differences. > For non-rgb, return the number of differing pixels." > | pixMask destShifted sourceShifted destPixVal bitsPerColor > rgbMask sourcePixVal diff maskShifted | > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > + <var: #sourceShifted type: 'unsigned int'> > + <var: #destShifted type: 'unsigned int'> > + <var: #maskShifted type: 'unsigned int'> > + <var: #pixMask type: 'unsigned int'> > + <var: #rgbMask type: 'unsigned int'> > pixMask := maskTable at: destDepth. > destDepth = 16 > ifTrue: [bitsPerColor := 5. rgbMask := 16r1F] > ifFalse: [bitsPerColor := 8. rgbMask := 16rFF]. > maskShifted := destMask. > destShifted := destinationWord. > sourceShifted := sourceWord. > 1 to: destPPW do: > [:i | > (maskShifted bitAnd: pixMask) > 0 ifTrue: > ["Only tally pixels within the destination > rectangle" > destPixVal := destShifted bitAnd: pixMask. > sourcePixVal := sourceShifted bitAnd: pixMask. > destDepth < 16 > ifTrue: [sourcePixVal = destPixVal > ifTrue: [diff := 0] > ifFalse: [diff := > 1]] > ifFalse: [diff := (self partitionedSub: > sourcePixVal from: destPixVal > nBits: > bitsPerColor nPartitions: 3). > diff := (diff bitAnd: > rgbMask) > + > (diff>>bitsPerColor bitAnd: rgbMask) > + > ((diff>>bitsPerColor)>>bitsPerColor bitAnd: rgbMask)]. > bitCount := bitCount + diff]. > maskShifted := maskShifted >> destDepth. > sourceShifted := sourceShifted >> destDepth. > destShifted := destShifted >> destDepth]. > ^ destinationWord "For no effect on dest" > ! > > Item was changed: > ----- Method: BitBltSimulation>>rgbMax:with: (in category 'combination > rules') ----- > rgbMax: sourceWord with: destinationWord > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > destDepth < 16 ifTrue: > ["Max each pixel separately" > ^ self partitionedMax: sourceWord with: destinationWord > nBits: destDepth > nPartitions: destPPW]. > destDepth = 16 ifTrue: > ["Max RGB components of each pixel separately" > ^ (self partitionedMax: sourceWord with: destinationWord > nBits: 5 nPartitions: 3) > + ((self partitionedMax: sourceWord>>16 with: > destinationWord>>16 > nBits: 5 nPartitions: 3) > << 16)] > ifFalse: > ["Max RGBA components of the pixel separately" > ^ self partitionedMax: sourceWord with: destinationWord > nBits: 8 nPartitions: 4]! > > Item was changed: > ----- Method: BitBltSimulation>>rgbMin:with: (in category 'combination > rules') ----- > rgbMin: sourceWord with: destinationWord > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > destDepth < 16 ifTrue: > ["Min each pixel separately" > ^ self partitionedMin: sourceWord with: destinationWord > nBits: destDepth > nPartitions: destPPW]. > destDepth = 16 ifTrue: > ["Min RGB components of each pixel separately" > ^ (self partitionedMin: sourceWord with: destinationWord > nBits: 5 nPartitions: 3) > + ((self partitionedMin: sourceWord>>16 with: > destinationWord>>16 > nBits: 5 nPartitions: 3) > << 16)] > ifFalse: > ["Min RGBA components of the pixel separately" > ^ self partitionedMin: sourceWord with: destinationWord > nBits: 8 nPartitions: 4]! > > Item was changed: > ----- Method: BitBltSimulation>>rgbMinInvert:with: (in category > 'combination rules') ----- > rgbMinInvert: wordToInvert with: destinationWord > | sourceWord | > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #wordToInvert type: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > sourceWord := wordToInvert bitInvert32. > destDepth < 16 ifTrue: > ["Min each pixel separately" > ^ self partitionedMin: sourceWord with: destinationWord > nBits: destDepth > nPartitions: destPPW]. > destDepth = 16 ifTrue: > ["Min RGB components of each pixel separately" > ^ (self partitionedMin: sourceWord with: destinationWord > nBits: 5 nPartitions: 3) > + ((self partitionedMin: sourceWord>>16 with: > destinationWord>>16 > nBits: 5 nPartitions: 3) > << 16)] > ifFalse: > ["Min RGBA components of the pixel separately" > ^ self partitionedMin: sourceWord with: destinationWord > nBits: 8 nPartitions: 4]! > > Item was changed: > ----- Method: BitBltSimulation>>rgbMul:with: (in category 'combination > rules') ----- > rgbMul: sourceWord with: destinationWord > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > destDepth < 16 ifTrue: > ["Mul each pixel separately" > ^ self partitionedMul: sourceWord with: destinationWord > nBits: destDepth > nPartitions: destPPW]. > destDepth = 16 ifTrue: > ["Mul RGB components of each pixel separately" > ^ (self partitionedMul: sourceWord with: destinationWord > nBits: 5 nPartitions: 3) > + ((self partitionedMul: sourceWord>>16 with: > destinationWord>>16 > nBits: 5 nPartitions: 3) > << 16)] > ifFalse: > ["Mul RGBA components of the pixel separately" > ^ self partitionedMul: sourceWord with: destinationWord > nBits: 8 nPartitions: 4] > > " | scanner | > Display repaintMorphicDisplay. > scanner := DisplayScanner quickPrintOn: Display. > MessageTally time: [0 to: 760 by: 4 do: [:y |scanner drawString: > 'qwrepoiuasfd=)(/&()=#!!〕kjzxv.,mn124+09857907QROIYTOAFDJZXNBNB,M-.,Mqwrepoiuasfd=)(/&()=#!!〕kjzxv.,mn124+09857907QROIYTOAFDJZXNBNB,M-.,M1234124356785678' > at: 0@y]]. "! > > Item was changed: > ----- Method: BitBltSimulation>>rgbSub:with: (in category 'combination > rules') ----- > rgbSub: sourceWord with: destinationWord > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > destDepth < 16 ifTrue: > ["Sub each pixel separately" > ^ self partitionedSub: sourceWord from: destinationWord > nBits: destDepth > nPartitions: destPPW]. > destDepth = 16 ifTrue: > ["Sub RGB components of each pixel separately" > ^ (self partitionedSub: sourceWord from: destinationWord > nBits: 5 nPartitions: 3) > + ((self partitionedSub: sourceWord>>16 from: > destinationWord>>16 > nBits: 5 nPartitions: 3) > << 16)] > ifFalse: > ["Sub RGBA components of the pixel separately" > ^ self partitionedSub: sourceWord from: destinationWord > nBits: 8 nPartitions: 4]! > > Item was changed: > ----- Method: BitBltSimulation>>sourceWord:with: (in category > 'combination rules') ----- > sourceWord: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord! > > Item was changed: > ----- Method: BitBltSimulation>>subWord:with: (in category 'combination > rules') ----- > subWord: sourceWord with: destinationWord > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > ^sourceWord - destinationWord! > > Item was changed: > ----- Method: BitBltSimulation>>tallyIntoMap:with: (in category > 'combination rules') ----- > tallyIntoMap: sourceWord with: destinationWord > "Tally pixels into the color map. Those tallied are exactly those > in the destination rectangle. Note that the source should be > specified == destination, in order for the proper color map checks > to be performed at setup." > | mapIndex pixMask destShifted maskShifted pixVal | > <inline: false> > + <returnTypeC: 'unsigned int'> > + <var: #sourceWord type: 'unsigned int'> > + <var: #destinationWord type: 'unsigned int'> > + <var: #pixMask type: 'unsigned int'> > + <var: #destShifted type: 'unsigned int'> > + <var: #maskShifted type: 'unsigned int'> > (cmFlags bitAnd: (ColorMapPresent bitOr: ColorMapIndexedPart)) = > (ColorMapPresent bitOr: ColorMapIndexedPart) > ifFalse: [^ destinationWord "no op"]. > pixMask := maskTable at: destDepth. > destShifted := destinationWord. > maskShifted := destMask. > 1 to: destPPW do: > [:i | > (maskShifted bitAnd: pixMask) = 0 ifFalse: > ["Only tally pixels within the destination > rectangle" > pixVal := destShifted bitAnd: pixMask. > destDepth < 16 > ifTrue: [mapIndex := pixVal] > ifFalse: [destDepth = 16 > ifTrue: [mapIndex := self rgbMap: > pixVal from: 5 to: cmBitsPerColor] > ifFalse: [mapIndex := self rgbMap: > pixVal from: 8 to: cmBitsPerColor]]. > self tallyMapAt: mapIndex put: (self tallyMapAt: > mapIndex) + 1]. > maskShifted := maskShifted >> destDepth. > destShifted := destShifted >> destDepth]. > ^ destinationWord "For no effect on dest"! > > Item was changed: > ----- Method: BitBltSimulation>>warpLoop (in category 'inner loop') ----- > warpLoop > "This version of the inner loop traverses an arbirary quadrilateral > source, thus producing a general affine transformation." > | skewWord halftoneWord mergeWord startBits > deltaP12x deltaP12y deltaP43x deltaP43y pAx pAy pBx pBy > xDelta yDelta smoothingCount sourceMapOop > nSteps nPix words destWord endBits mergeFnwith dstShiftInc > dstShiftLeft mapperFlags | > + <inline: false> > + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned > int, unsigned int)'> > + <var: #skewWord type: #'unsigned int'> > + <var: #halftoneWord type: #'unsigned int'> > + <var: #mergeWord type: #'unsigned int'> > + <var: #destWord type: #'unsigned int'> > + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'unsigned int (*)(unsigned int, unsigned int)'. > - <inline: false> > - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> > - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: > 'sqInt (*)(sqInt, sqInt)'. > mergeFnwith. "null ref for compiler" > > (interpreterProxy slotSizeOf: bitBltOop) >= (BBWarpBase+12) > ifFalse: [^ interpreterProxy primitiveFail]. > nSteps := height-1. nSteps <= 0 ifTrue: [nSteps := 1]. > > pAx := self fetchIntOrFloat: BBWarpBase ofObject: bitBltOop. > words := self fetchIntOrFloat: BBWarpBase+3 ofObject: bitBltOop. > deltaP12x := self deltaFrom: pAx to: words nSteps: nSteps. > deltaP12x < 0 ifTrue: [pAx := words - (nSteps*deltaP12x)]. > > pAy := self fetchIntOrFloat: BBWarpBase+1 ofObject: bitBltOop. > words := self fetchIntOrFloat: BBWarpBase+4 ofObject: bitBltOop. > deltaP12y := self deltaFrom: pAy to: words nSteps: nSteps. > deltaP12y < 0 ifTrue: [pAy := words - (nSteps*deltaP12y)]. > > pBx := self fetchIntOrFloat: BBWarpBase+9 ofObject: bitBltOop. > words := self fetchIntOrFloat: BBWarpBase+6 ofObject: bitBltOop. > deltaP43x := self deltaFrom: pBx to: words nSteps: nSteps. > deltaP43x < 0 ifTrue: [pBx := words - (nSteps*deltaP43x)]. > > pBy := self fetchIntOrFloat: BBWarpBase+10 ofObject: bitBltOop. > words := self fetchIntOrFloat: BBWarpBase+7 ofObject: bitBltOop. > deltaP43y := self deltaFrom: pBy to: words nSteps: nSteps. > deltaP43y < 0 ifTrue: [pBy := words - (nSteps*deltaP43y)]. > > interpreterProxy failed ifTrue: [^ false]. "ie if non-integers > above" > interpreterProxy methodArgumentCount = 2 > ifTrue: [smoothingCount := interpreterProxy > stackIntegerValue: 1. > sourceMapOop := interpreterProxy > stackValue: 0. > sourceMapOop = interpreterProxy nilObject > ifTrue: [sourceDepth < 16 ifTrue: > ["color map is required to smooth > non-RGB dest" > ^ interpreterProxy primitiveFail]] > ifFalse: [(interpreterProxy slotSizeOf: > sourceMapOop) > < (1 << > sourceDepth) ifTrue: > ["sourceMap must be long enough > for sourceDepth" > ^ interpreterProxy primitiveFail]. > sourceMapOop := self > oopForPointer: (interpreterProxy firstIndexableField: sourceMapOop)]] > ifFalse: [smoothingCount := 1. > sourceMapOop := interpreterProxy > nilObject]. > nSteps := width-1. nSteps <= 0 ifTrue: [nSteps := 1]. > startBits := destPPW - (dx bitAnd: destPPW-1). > endBits := ((dx + bbW - 1) bitAnd: destPPW-1) + 1. > bbW < startBits ifTrue:[startBits := bbW]. > > destY < clipY ifTrue:[ > "Advance increments if there was clipping in y" > pAx := pAx + (clipY - destY * deltaP12x). > pAy := pAy + (clipY - destY * deltaP12y). > pBx := pBx + (clipY - destY * deltaP43x). > pBy := pBy + (clipY - destY * deltaP43y)]. > > "Setup values for faster pixel fetching." > self warpLoopSetup. > "Setup color mapping if not provided" > (smoothingCount > 1 and:[(cmFlags bitAnd: ColorMapNewStyle) = 0]) > ifTrue:[ > cmLookupTable == nil ifTrue:[ > destDepth = 16 ifTrue:[self setupColorMasksFrom: 8 > to: 5]. > ] ifFalse:[ > self setupColorMasksFrom: 8 to: cmBitsPerColor. > ]. > ]. > mapperFlags := cmFlags bitAnd: ColorMapNewStyle bitInvert32. > > destMSB > ifTrue:[ dstShiftInc := 0 - destDepth. > dstShiftLeft := 32 - destDepth] > ifFalse:[ dstShiftInc := destDepth. > dstShiftLeft := 0]. > 1 to: bbH do: > [ :i | "here is the vertical loop..." > xDelta := self deltaFrom: pAx to: pBx nSteps: nSteps. > xDelta >= 0 ifTrue: [sx := pAx] ifFalse: [sx := pBx - > (nSteps*xDelta)]. > yDelta := self deltaFrom: pAy to: pBy nSteps: nSteps. > yDelta >= 0 ifTrue: [sy := pAy] ifFalse: [sy := pBy - > (nSteps*yDelta)]. > > destMSB > ifTrue:[dstBitShift := 32 - ((dx bitAnd: destPPW - > 1) + 1 * destDepth)] > ifFalse:[dstBitShift := (dx bitAnd: destPPW - 1) * > destDepth]. > > (destX < clipX) ifTrue:[ > "Advance increments if there was clipping in x" > sx := sx + (clipX - destX * xDelta). > sy := sy + (clipX - destX * yDelta). > ]. > > noHalftone > ifTrue: [halftoneWord := AllOnes] > ifFalse: [halftoneWord := self halftoneAt: dy+i-1]. > destMask := mask1. > nPix := startBits. > "Here is the inner loop..." > words := nWords. > ["pick up word" > smoothingCount = 1 ifTrue:["Faster if not > smoothing" > skewWord := self warpPickSourcePixels: nPix > xDeltah: > xDelta yDeltah: yDelta > xDeltav: > deltaP12x yDeltav: deltaP12y > > dstShiftInc: dstShiftInc flags: mapperFlags. > ] ifFalse:["more difficult with smoothing" > skewWord := self warpPickSmoothPixels: nPix > xDeltah: xDelta yDeltah: > yDelta > xDeltav: deltaP12x > yDeltav: deltaP12y > sourceMap: sourceMapOop > smoothing: smoothingCount > dstShiftInc: dstShiftInc. > ]. > "align next word access to left most pixel" > dstBitShift := dstShiftLeft. > destMask = AllOnes ifTrue:["avoid > read-modify-write" > mergeWord := self mergeFn: (skewWord > bitAnd: halftoneWord) > with: > (self dstLongAt: destIndex). > self dstLongAt: destIndex put: (destMask > bitAnd: mergeWord). > ] ifFalse:[ "General version using dest masking" > destWord := self dstLongAt: destIndex. > mergeWord := self mergeFn: (skewWord > bitAnd: halftoneWord) > with: > (destWord bitAnd: destMask). > destWord := (destMask bitAnd: mergeWord) > bitOr: > (destWord > bitAnd: destMask bitInvert32). > self dstLongAt: destIndex put: destWord. > ]. > self incDestIndex: 4. > words = 2 "e.g., is the next word the last word?" > ifTrue:["set mask for last word in this > row" > destMask := mask2. > nPix := endBits] > ifFalse:["use fullword mask for inner loop" > destMask := AllOnes. > nPix := destPPW]. > (words := words - 1) = 0] whileFalse. > "--- end of inner loop ---" > pAx := pAx + deltaP12x. > pAy := pAy + deltaP12y. > pBx := pBx + deltaP43x. > pBy := pBy + deltaP43y. > self incDestIndex: destDelta]! > > Item was changed: > ----- Method: > BitBltSimulation>>warpPickSmoothPixels:xDeltah:yDeltah:xDeltav:yDeltav:sourceMap:smoothing:dstShiftInc: > (in category 'pixel mapping') ----- > warpPickSmoothPixels: nPixels > xDeltah: xDeltah yDeltah: yDeltah > xDeltav: xDeltav yDeltav: yDeltav > sourceMap: sourceMap > smoothing: n > dstShiftInc: dstShiftInc > "Pick n (sub-) pixels from the source form, mapped by sourceMap, > average the RGB values, map by colorMap and return the new word. > This version is only called from WarpBlt with smoothingCount > 1" > | rgb x y a r g b xx yy xdh ydh xdv ydv dstMask destWord i j k > nPix | > <inline: false> "nope - too much stuff in here" > + <var: #rgb type: #'unsigned int'> > dstMask := maskTable at: destDepth. > destWord := 0. > n = 2 "Try avoiding divides for most common n (divide by 2 is > generated as shift)" > ifTrue:[xdh := xDeltah // 2. ydh := yDeltah // 2. > xdv := xDeltav // 2. ydv := yDeltav // 2] > ifFalse:[xdh := xDeltah // n. ydh := yDeltah // n. > xdv := xDeltav // n. ydv := yDeltav // n]. > i := nPixels. > [ > x := sx. y := sy. > a := r := g := b := 0. > "Pick and average n*n subpixels" > nPix := 0. "actual number of pixels (not clipped and not > transparent)" > j := n. > [ > xx := x. yy := y. > k := n. > [ > "get a single subpixel" > rgb := self pickWarpPixelAtX: xx y: yy. > (combinationRule=25 "PAINT" and: [rgb = > 0]) ifFalse:[ > "If not clipped and not > transparent, then tally rgb values" > nPix := nPix + 1. > sourceDepth < 16 ifTrue:[ > "Get RGBA values from > sourcemap table" > rgb := self long32At: > sourceMap + (rgb << 2). > ] ifFalse:["Already in RGB format" > sourceDepth = 16 > > ifTrue:[rgb := self rgbMap16To32: rgb] > > ifFalse:[rgb := self rgbMap32To32: rgb]]. > b := b + (rgb bitAnd: 255). > g := g + (rgb >> 8 bitAnd: 255). > r := r + (rgb >> 16 bitAnd: 255). > a := a + (rgb >> 24)]. > xx := xx + xdh. > yy := yy + ydh. > (k := k - 1) = 0] whileFalse. > x := x + xdv. > y := y + ydv. > (j := j - 1) = 0] whileFalse. > > (nPix = 0 or: [combinationRule=25 "PAINT" and: [nPix < (n > * n // 2)]]) ifTrue:[ > rgb := 0 "All pixels were 0, or most were > transparent" > ] ifFalse:[ > "normalize rgba sums" > nPix = 4 "Try to avoid divides for most common n" > ifTrue:[r := r >> 2. g := g >> 2. b > := b >> 2. a := a >> 2] > ifFalse:[ r := r // nPix. g := g // > nPix. b := b // nPix. a := a // nPix]. > rgb := (a << 24) + (r << 16) + (g << 8) + b. > > "map the pixel" > rgb = 0 ifTrue: [ > "only generate zero if pixel is really > transparent" > (r + g + b + a) > 0 ifTrue: [rgb := 1]]. > rgb := self mapPixel: rgb flags: cmFlags. > ]. > "Mix it in" > destWord := destWord bitOr: (rgb bitAnd: dstMask) << > dstBitShift. > dstBitShift := dstBitShift + dstShiftInc. > sx := sx + xDeltah. > sy := sy + yDeltah. > (i := i - 1) = 0] whileFalse. > > ^destWord > ! > >

2 1

VM Maker: VMMaker.oscog-nice.1754.mcz
by commits＠source.squeak.org 31 Mar '16

31 Mar '16

Nicolas Cellier uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-nice.1754.mcz ==================== Summary ==================== Name: VMMaker.oscog-nice.1754 Author: nice Time: 31 March 2016, 10:04:44.274 pm UUID: a0138c60-3e62-446e-a1d8-b00b7bb6c92b Ancestors: VMMaker.oscog-eem.1753 BitBltSimulation is operating on bits of 32bits words, therefore it's better to declare its operans as 'unsigned int' rather than sqInt. On 32bits VM, this should not change anything, but on 64bits spur, it makes this snippet work: | wideString source pos blt expectedWideString | source := #[1 64 255 14 1 64 48 251]. expectedWideString := WideString fromByteArray: source. wideString := WideString new: source size // 4. pos := 0. blt := (BitBlt toForm: (Form new hackBits: wideString)) sourceForm: (Form new hackBits: source). blt combinationRule: Form over; sourceX: 0; sourceY: pos // 4; height: wideString byteSize // 4; width: 4; destX: 0; destY: 0; copyBits. wideString restoreEndianness. self assert: wideString = expectedWideString Hence it fix loading/diffing MCZ with wide character. =============== Diff against VMMaker.oscog-eem.1753 =============== Item was changed: ----- Method: BitBltSimulation>>addWord:with: (in category 'combination rules') ----- addWord: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord + destinationWord! Item was changed: ----- Method: BitBltSimulation>>alphaBlend:with: (in category 'combination rules') ----- alphaBlend: sourceWord with: destinationWord "Blend sourceWord with destinationWord, assuming both are 32-bit pixels. The source is assumed to have 255*alpha in the high 8 bits of each pixel, while the high 8 bits of the destinationWord will be ignored. The blend produced is alpha*source + (1-alpha)*dest, with the computation being performed independently on each color component. The high byte of the result will be 0." | alpha unAlpha result blendRB blendAG | <inline: false> + <returnTypeC: 'unsigned int'> - <return: 'unsigned int'> <var: #sourceWord type: 'unsigned int'> <var: #destinationWord type: 'unsigned int'> <var: #blendRB type: 'unsigned int'> <var: #blendAG type: 'unsigned int'> <var: #result type: 'unsigned int'> <var: #alpha type: 'unsigned int'> <var: #unAlpha type: 'unsigned int'> alpha := sourceWord >> 24. "High 8 bits of source pixel" alpha = 0 ifTrue: [ ^ destinationWord ]. alpha = 255 ifTrue: [ ^ sourceWord ]. unAlpha := 255 - alpha. blendRB := ((sourceWord bitAnd: 16rFF00FF) * alpha) + ((destinationWord bitAnd: 16rFF00FF) * unAlpha) + 16rFF00FF. "blend red and blue" blendAG := (((sourceWord>> 8 bitOr: 16rFF0000) bitAnd: 16rFF00FF) * alpha) + ((destinationWord>>8 bitAnd: 16rFF00FF) * unAlpha) + 16rFF00FF. "blend alpha and green" blendRB := blendRB + (blendRB - 16r10001 >> 8 bitAnd: 16rFF00FF) >> 8 bitAnd: 16rFF00FF. "divide by 255" blendAG := blendAG + (blendAG - 16r10001 >> 8 bitAnd: 16rFF00FF) >> 8 bitAnd: 16rFF00FF. result := blendRB bitOr: blendAG<<8. ^ result ! Item was changed: ----- Method: BitBltSimulation>>alphaBlendConst:with: (in category 'combination rules') ----- alphaBlendConst: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^ self alphaBlendConst: sourceWord with: destinationWord paintMode: false! Item was changed: ----- Method: BitBltSimulation>>alphaBlendConst:with:paintMode: (in category 'combination rules') ----- alphaBlendConst: sourceWord with: destinationWord paintMode: paintMode "Blend sourceWord with destinationWord using a constant alpha. Alpha is encoded as 0 meaning 0.0, and 255 meaning 1.0. The blend produced is alpha*source + (1.0-alpha)*dest, with the computation being performed independently on each color component. This function could eventually blend into any depth destination, using the same color averaging and mapping as warpBlt. paintMode = true means do nothing if the source pixel value is zero." "This first implementation works with dest depths of 16 and 32 bits only. Normal color mapping will allow sources of lower depths in this case, and results can be mapped directly by truncation, so no extra color maps are needed. To allow storing into any depth will require subsequent addition of two other colormaps, as is the case with WarpBlt." | pixMask destShifted sourceShifted destPixVal rgbMask sourcePixVal unAlpha result pixBlend shift blend maskShifted bitsPerColor blendAG blendRB | <inline: false> + <returnTypeC: 'unsigned int'> - <return: 'unsigned int'> <var: #sourceWord type: 'unsigned int'> <var: #destinationWord type: 'unsigned int'> <var: #blendRB type: 'unsigned int'> <var: #blendAG type: 'unsigned int'> <var: #result type: 'unsigned int'> <var: #sourceAlpha type: 'unsigned int'> <var: #unAlpha type: 'unsigned int'> + <var: #sourceShifted type: 'unsigned int'> + <var: #destShifted type: 'unsigned int'> + <var: #maskShifted type: 'unsigned int'> + <var: #pixMask type: 'unsigned int'> + <var: #rgbMask type: 'unsigned int'> + <var: #pixBlend type: 'unsigned int'> + <var: #blend type: 'unsigned int'> destDepth < 16 ifTrue: [^ destinationWord "no-op"]. unAlpha := 255 - sourceAlpha. result := destinationWord. destPPW = 1 ifTrue:["32bpp blends include alpha" paintMode & (sourceWord = 0) "painting a transparent pixel" ifFalse:[ blendRB := ((sourceWord bitAnd: 16rFF00FF) * sourceAlpha) + ((destinationWord bitAnd: 16rFF00FF) * unAlpha) + 16rFF00FF. "blendRB red and blue" blendAG := ((sourceWord>> 8 bitAnd: 16rFF00FF) * sourceAlpha) + ((destinationWord>>8 bitAnd: 16rFF00FF) * unAlpha) + 16rFF00FF. "blendRB alpha and green" blendRB := blendRB + (blendRB - 16r10001 >> 8 bitAnd: 16rFF00FF) >> 8 bitAnd: 16rFF00FF. "divide by 255" blendAG := blendAG + (blendAG - 16r10001 >> 8 bitAnd: 16rFF00FF) >> 8 bitAnd: 16rFF00FF. result := blendRB bitOr: blendAG<<8. ]. ] ifFalse:[ pixMask := maskTable at: destDepth. bitsPerColor := 5. rgbMask := 16r1F. maskShifted := destMask. destShifted := destinationWord. sourceShifted := sourceWord. 1 to: destPPW do:[:j | sourcePixVal := sourceShifted bitAnd: pixMask. ((maskShifted bitAnd: pixMask) = 0 "no effect if outside of dest rectangle" or: [paintMode & (sourcePixVal = 0) "or painting a transparent pixel"]) ifFalse: [destPixVal := destShifted bitAnd: pixMask. pixBlend := 0. 1 to: 3 do: [:i | shift := (i-1)*bitsPerColor. blend := (((sourcePixVal>>shift bitAnd: rgbMask) * sourceAlpha) + ((destPixVal>>shift bitAnd: rgbMask) * unAlpha)) + 254 // 255 bitAnd: rgbMask. pixBlend := pixBlend bitOr: blend<<shift]. result := (result bitAnd: (pixMask << (j-1*16)) bitInvert32) bitOr: pixBlend << (j-1*16)]. maskShifted := maskShifted >> destDepth. sourceShifted := sourceShifted >> destDepth. destShifted := destShifted >> destDepth]. ]. ^ result ! Item was changed: ----- Method: BitBltSimulation>>alphaBlendScaled:with: (in category 'combination rules') ----- alphaBlendScaled: sourceWord with: destinationWord "Blend sourceWord with destinationWord using the alpha value from sourceWord. Alpha is encoded as 0 meaning 0.0, and 255 meaning 1.0. In contrast to alphaBlend:with: the color produced is srcColor + (1-srcAlpha) * dstColor e.g., it is assumed that the source color is already scaled." | unAlpha rb ag | <inline: false> "Do NOT inline this into optimized loops" + <returnTypeC: 'unsigned int'> <var: #sourceWord type: 'unsigned int'> <var: #destinationWord type: 'unsigned int'> <var: #rb type: 'unsigned int'> <var: #ag type: 'unsigned int'> <var: #unAlpha type: 'unsigned int'> unAlpha := 255 - (sourceWord >> 24). "High 8 bits of source pixel is source opacity (ARGB format)" rb := ((destinationWord bitAnd: 16rFF00FF) * unAlpha >> 8 bitAnd: 16rFF00FF) + (sourceWord bitAnd: 16rFF00FF). "blend red and blue components" ag := ((destinationWord>>8 bitAnd: 16rFF00FF) * unAlpha >> 8 bitAnd: 16rFF00FF) + (sourceWord>>8 bitAnd: 16rFF00FF). "blend alpha and green components" rb := (rb bitAnd: 16rFF00FF) bitOr: (rb bitAnd: 16r1000100) * 16rFF >> 8. "saturate red and blue components if there is a carry" ag := (ag bitAnd: 16rFF00FF) << 8 bitOr: (ag bitAnd: 16r1000100) * 16rFF. "saturate alpha and green components if there is a carry" ^ag bitOr: rb "recompose"! Item was changed: ----- Method: BitBltSimulation>>alphaPaintConst:with: (in category 'combination rules') ----- alphaPaintConst: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> sourceWord = 0 ifTrue: [^ destinationWord "opt for all-transparent source"]. ^ self alphaBlendConst: sourceWord with: destinationWord paintMode: true! Item was changed: ----- Method: BitBltSimulation>>alphaSourceBlendBits16 (in category 'inner loop') ----- alphaSourceBlendBits16 "This version assumes combinationRule = 34 sourcePixSize = 32 destPixSize = 16 sourceForm ~= destForm. " | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY srcY dstY dstMask srcShift ditherBase ditherIndex ditherThreshold | <inline: false> "This particular method should be optimized in itself" + <var: #sourceWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> + <var: #dstMask type: #'unsigned int'> deltaY := bbH + 1. "So we can pre-decrement" srcY := sy. dstY := dy. srcShift := (dx bitAnd: 1) * 16. destMSB ifTrue:[srcShift := 16 - srcShift]. mask1 := 16rFFFF << (16 - srcShift). "This is the outer loop" [(deltaY := deltaY - 1) ~= 0] whileTrue:[ srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). dstIndex := destBits + (dstY * destPitch) + (dx // 2 * 4). ditherBase := (dstY bitAnd: 3) * 4. ditherIndex := (sx bitAnd: 3) - 1. "For pre-increment" deltaX := bbW + 1. "So we can pre-decrement" dstMask := mask1. dstMask = 16rFFFF ifTrue:[srcShift := 16] ifFalse:[srcShift := 0]. "This is the inner loop" [(deltaX := deltaX - 1) ~= 0] whileTrue:[ ditherThreshold := ditherMatrix4x4 at: ditherBase + (ditherIndex := ditherIndex + 1 bitAnd: 3). sourceWord := self srcLongAt: srcIndex. srcAlpha := sourceWord >> 24. srcAlpha = 255 ifTrue:[ "Dither from 32 to 16 bit" sourceWord := self dither32To16: sourceWord threshold: ditherThreshold. sourceWord = 0 ifTrue:[sourceWord := 1 << srcShift] ifFalse: [sourceWord := sourceWord << srcShift]. "Store masked value" self dstLongAt: dstIndex put: sourceWord mask: dstMask. ] ifFalse:[ "srcAlpha ~= 255" srcAlpha = 0 ifFalse:[ "0 < srcAlpha < 255" "If we have to mix colors then just copy a single word" destWord := self dstLongAt: dstIndex. destWord := destWord bitAnd: dstMask bitInvert32. destWord := destWord >> srcShift. "Expand from 16 to 32 bit by adding zero bits" destWord := (((destWord bitAnd: 16r7C00) bitShift: 9) bitOr: ((destWord bitAnd: 16r3E0) bitShift: 6)) bitOr: (((destWord bitAnd: 16r1F) bitShift: 3) bitOr: 16rFF000000). "Mix colors" sourceWord := self alphaBlendScaled: sourceWord with: destWord. "And dither" sourceWord := self dither32To16: sourceWord threshold: ditherThreshold. sourceWord = 0 ifTrue:[sourceWord := 1 << srcShift] ifFalse:[sourceWord := sourceWord << srcShift]. "Store back" self dstLongAt: dstIndex put: sourceWord mask: dstMask. ]. ]. srcIndex := srcIndex + 4. destMSB ifTrue:[srcShift = 0 ifTrue:[dstIndex := dstIndex + 4]] ifFalse:[srcShift = 0 ifFalse:[dstIndex := dstIndex + 4]]. srcShift := srcShift bitXor: 16. "Toggle between 0 and 16" dstMask := dstMask bitInvert32. "Mask other half word" ]. srcY := srcY + 1. dstY := dstY + 1. ].! Item was changed: ----- Method: BitBltSimulation>>alphaSourceBlendBits32 (in category 'inner loop') ----- alphaSourceBlendBits32 "This version assumes combinationRule = 34 sourcePixSize = destPixSize = 32 sourceForm ~= destForm. Note: The inner loop has been optimized for dealing with the special cases of srcAlpha = 0.0 and srcAlpha = 1.0 " | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY srcY dstY | <inline: false> "This particular method should be optimized in itself" + <var: #sourceWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> "Give the compile a couple of hints" "The following should be declared as pointers so the compiler will notice that they're used for accessing memory locations (good to know on an Intel architecture) but then the increments would be different between ST code and C code so must hope the compiler notices what happens (MS Visual C does)" deltaY := bbH + 1. "So we can pre-decrement" srcY := sy. dstY := dy. "This is the outer loop" [(deltaY := deltaY - 1) ~= 0] whileTrue:[ srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). dstIndex := destBits + (dstY * destPitch) + (dx * 4). deltaX := bbW + 1. "So we can pre-decrement" "This is the inner loop" [(deltaX := deltaX - 1) ~= 0] whileTrue:[ sourceWord := self srcLongAt: srcIndex. srcAlpha := sourceWord >> 24. srcAlpha = 255 ifTrue:[ self dstLongAt: dstIndex put: sourceWord. srcIndex := srcIndex + 4. dstIndex := dstIndex + 4. "Now copy as many words as possible with alpha = 255" [(deltaX := deltaX - 1) ~= 0 and:[ (sourceWord := self srcLongAt: srcIndex) >> 24 = 255]] whileTrue:[ self dstLongAt: dstIndex put: sourceWord. srcIndex := srcIndex + 4. dstIndex := dstIndex + 4. ]. "Adjust deltaX" deltaX := deltaX + 1. ] ifFalse:[ "srcAlpha ~= 255" srcAlpha = 0 ifTrue:[ srcIndex := srcIndex + 4. dstIndex := dstIndex + 4. "Now skip as many words as possible," [(deltaX := deltaX - 1) ~= 0 and:[ (sourceWord := self srcLongAt: srcIndex) >> 24 = 0]] whileTrue:[ srcIndex := srcIndex + 4. dstIndex := dstIndex + 4. ]. "Adjust deltaX" deltaX := deltaX + 1. ] ifFalse:[ "0 < srcAlpha < 255" "If we have to mix colors then just copy a single word" destWord := self dstLongAt: dstIndex. destWord := self alphaBlendScaled: sourceWord with: destWord. self dstLongAt: dstIndex put: destWord. srcIndex := srcIndex + 4. dstIndex := dstIndex + 4. ]. ]. ]. srcY := srcY + 1. dstY := dstY + 1. ].! Item was changed: ----- Method: BitBltSimulation>>alphaSourceBlendBits8 (in category 'inner loop') ----- alphaSourceBlendBits8 "This version assumes combinationRule = 34 sourcePixSize = 32 destPixSize = 8 sourceForm ~= destForm. Note: This is not real blending since we don't have the source colors available. " | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY srcY dstY dstMask srcShift adjust mappingTable mapperFlags | <inline: false> <var: #mappingTable type:'unsigned int *'> + <var: #sourceWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> + <var: #dstMask type: #'unsigned int'> mappingTable := self default8To32Table. mapperFlags := cmFlags bitAnd: ColorMapNewStyle bitInvert32. deltaY := bbH + 1. "So we can pre-decrement" srcY := sy. dstY := dy. mask1 := ((dx bitAnd: 3) * 8). destMSB ifTrue:[mask1 := 24 - mask1]. mask2 := AllOnes bitXor:(16rFF << mask1). (dx bitAnd: 1) = 0 ifTrue:[adjust := 0] ifFalse:[adjust := 16r1F1F1F1F]. (dy bitAnd: 1) = 0 ifTrue:[adjust := adjust bitXor: 16r1F1F1F1F]. "This is the outer loop" [(deltaY := deltaY - 1) ~= 0] whileTrue:[ adjust := adjust bitXor: 16r1F1F1F1F. srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). dstIndex := destBits + (dstY * destPitch) + (dx // 4 * 4). deltaX := bbW + 1. "So we can pre-decrement" srcShift := mask1. dstMask := mask2. "This is the inner loop" [(deltaX := deltaX - 1) ~= 0] whileTrue:[ sourceWord := ((self srcLongAt: srcIndex) bitAnd: (adjust bitInvert32)) + adjust. srcAlpha := sourceWord >> 24. srcAlpha > 31 ifTrue:["Everything below 31 is transparent" srcAlpha < 224 ifTrue:["Everything above 224 is opaque" destWord := self dstLongAt: dstIndex. destWord := destWord bitAnd: dstMask bitInvert32. destWord := destWord >> srcShift. destWord := mappingTable at: destWord. sourceWord := self alphaBlendScaled: sourceWord with: destWord. ]. sourceWord := self mapPixel: sourceWord flags: mapperFlags. sourceWord := sourceWord << srcShift. "Store back" self dstLongAt: dstIndex put: sourceWord mask: dstMask. ]. srcIndex := srcIndex + 4. destMSB ifTrue:[ srcShift = 0 ifTrue:[dstIndex := dstIndex + 4. srcShift := 24. dstMask := 16r00FFFFFF] ifFalse:[srcShift := srcShift - 8. dstMask := (dstMask >> 8) bitOr: 16rFF000000]. ] ifFalse:[ srcShift = 24 ifTrue:[dstIndex := dstIndex + 4. srcShift := 0. dstMask := 16rFFFFFF00] ifFalse:[srcShift := srcShift + 8. dstMask := dstMask << 8 bitOr: 255]. ]. adjust := adjust bitXor: 16r1F1F1F1F. ]. srcY := srcY + 1. dstY := dstY + 1. ].! Item was changed: ----- Method: BitBltSimulation>>bitAnd:with: (in category 'combination rules') ----- bitAnd: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitAnd: destinationWord! Item was changed: ----- Method: BitBltSimulation>>bitAndInvert:with: (in category 'combination rules') ----- bitAndInvert: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitAnd: destinationWord bitInvert32! Item was changed: ----- Method: BitBltSimulation>>bitInvertAnd:with: (in category 'combination rules') ----- bitInvertAnd: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitInvert32 bitAnd: destinationWord! Item was changed: ----- Method: BitBltSimulation>>bitInvertAndInvert:with: (in category 'combination rules') ----- bitInvertAndInvert: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitInvert32 bitAnd: destinationWord bitInvert32! Item was changed: ----- Method: BitBltSimulation>>bitInvertDestination:with: (in category 'combination rules') ----- bitInvertDestination: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^destinationWord bitInvert32! Item was changed: ----- Method: BitBltSimulation>>bitInvertOr:with: (in category 'combination rules') ----- bitInvertOr: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitInvert32 bitOr: destinationWord! Item was changed: ----- Method: BitBltSimulation>>bitInvertOrInvert:with: (in category 'combination rules') ----- bitInvertOrInvert: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitInvert32 bitOr: destinationWord bitInvert32! Item was changed: ----- Method: BitBltSimulation>>bitInvertSource:with: (in category 'combination rules') ----- bitInvertSource: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitInvert32! Item was changed: ----- Method: BitBltSimulation>>bitInvertXor:with: (in category 'combination rules') ----- bitInvertXor: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitInvert32 bitXor: destinationWord! Item was changed: ----- Method: BitBltSimulation>>bitOr:with: (in category 'combination rules') ----- bitOr: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitOr: destinationWord! Item was changed: ----- Method: BitBltSimulation>>bitOrInvert:with: (in category 'combination rules') ----- bitOrInvert: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitOr: destinationWord bitInvert32! Item was changed: ----- Method: BitBltSimulation>>bitXor:with: (in category 'combination rules') ----- bitXor: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord bitXor: destinationWord! Item was changed: ----- Method: BitBltSimulation>>clearWord:with: (in category 'combination rules') ----- + clearWord: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> - clearWord: source with: destination ^ 0! Item was changed: ----- Method: BitBltSimulation>>copyLoop (in category 'inner loop') ----- copyLoop | prevWord thisWord skewWord halftoneWord mergeWord hInc y unskew skewMask notSkewMask mergeFnwith destWord | "This version of the inner loop assumes noSource = false." <inline: false> + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned int, unsigned int)'> + <var: #prevWord type: #'unsigned int'> + <var: #thisWord type: #'unsigned int'> + <var: #skewWord type: #'unsigned int'> + <var: #halftoneWord type: #'unsigned int'> + <var: #mergeWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> + <var: #skewMask type: #'unsigned int'> + <var: #notSkewMask type: #'unsigned int'> + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'unsigned int (*)(unsigned int, unsigned int)'. - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'sqInt (*)(sqInt, sqInt)'. mergeFnwith. "null ref for compiler" hInc := hDir*4. "Byte delta" "degenerate skew fixed for Sparc. 10/20/96 ikp" skew == -32 ifTrue: [skew := unskew := skewMask := 0] ifFalse: [skew < 0 ifTrue: [unskew := skew+32. skewMask := AllOnes << (0-skew)] ifFalse: [skew = 0 ifTrue: [unskew := 0. skewMask := AllOnes] ifFalse: [unskew := skew-32. skewMask := AllOnes >> skew]]]. notSkewMask := skewMask bitInvert32. noHalftone ifTrue: [halftoneWord := AllOnes. halftoneHeight := 0] ifFalse: [halftoneWord := self halftoneAt: 0]. y := dy. 1 to: bbH do: "here is the vertical loop" [ :i | halftoneHeight > 1 ifTrue: "Otherwise, its always the same" [halftoneWord := self halftoneAt: y. y := y + vDir]. preload ifTrue: ["load the 64-bit shifter" prevWord := self srcLongAt: sourceIndex. self incSrcIndex: hInc] ifFalse: [prevWord := 0]. "Note: the horizontal loop has been expanded into three parts for speed:" "This first section requires masking of the destination store..." destMask := mask1. thisWord := self srcLongAt: sourceIndex. "pick up next word" self incSrcIndex: hInc. skewWord := ((prevWord bitAnd: notSkewMask) bitShift: unskew) bitOr: "32-bit rotate" ((thisWord bitAnd: skewMask) bitShift: skew). prevWord := thisWord. destWord := self dstLongAt: destIndex. mergeWord := self mergeFn: (skewWord bitAnd: halftoneWord) with: destWord. destWord := (destMask bitAnd: mergeWord) bitOr: (destWord bitAnd: destMask bitInvert32). self dstLongAt: destIndex put: destWord. self incDestIndex: hInc. "This central horizontal loop requires no store masking" destMask := AllOnes. combinationRule = 3 ifTrue: [(skew = 0) & (halftoneWord = AllOnes) ifTrue: ["Very special inner loop for STORE mode with no skew -- just move words" hDir = -1 ifTrue: ["Woeful patch: revert to older code for hDir = -1" 2 to: nWords-1 do: [ :word | thisWord := self srcLongAt: sourceIndex. self incSrcIndex: hInc. self dstLongAt: destIndex put: thisWord. self incDestIndex: hInc]] ifFalse: [2 to: nWords-1 do: [ :word | "Note loop starts with prevWord loaded (due to preload)" self dstLongAt: destIndex put: prevWord. self incDestIndex: hInc. prevWord := self srcLongAt: sourceIndex. self incSrcIndex: hInc]]] ifFalse: ["Special inner loop for STORE mode -- no need to call merge" 2 to: nWords-1 do: [ :word | thisWord := self srcLongAt: sourceIndex. self incSrcIndex: hInc. skewWord := ((prevWord bitAnd: notSkewMask) bitShift: unskew) bitOr: "32-bit rotate" ((thisWord bitAnd: skewMask) bitShift: skew). prevWord := thisWord. self dstLongAt: destIndex put: (skewWord bitAnd: halftoneWord). self incDestIndex: hInc]] ] ifFalse: [2 to: nWords-1 do: "Normal inner loop does merge:" [ :word | thisWord := self srcLongAt: sourceIndex. "pick up next word" self incSrcIndex: hInc. skewWord := ((prevWord bitAnd: notSkewMask) bitShift: unskew) bitOr: "32-bit rotate" ((thisWord bitAnd: skewMask) bitShift: skew). prevWord := thisWord. mergeWord := self mergeFn: (skewWord bitAnd: halftoneWord) with: (self dstLongAt: destIndex). self dstLongAt: destIndex put: mergeWord. self incDestIndex: hInc] ]. "This last section, if used, requires masking of the destination store..." nWords > 1 ifTrue: [destMask := mask2. thisWord := self srcLongAt: sourceIndex. "pick up next word" self incSrcIndex: hInc. skewWord := ((prevWord bitAnd: notSkewMask) bitShift: unskew) bitOr: "32-bit rotate" ((thisWord bitAnd: skewMask) bitShift: skew). destWord := self dstLongAt: destIndex. mergeWord := self mergeFn: (skewWord bitAnd: halftoneWord) with: destWord. destWord := (destMask bitAnd: mergeWord) bitOr: (destWord bitAnd: destMask bitInvert32). self dstLongAt: destIndex put: destWord. self incDestIndex: hInc]. self incSrcIndex: sourceDelta. self incDestIndex: destDelta]! Item was changed: ----- Method: BitBltSimulation>>copyLoopNoSource (in category 'inner loop') ----- copyLoopNoSource "Faster copyLoop when source not used. hDir and vDir are both positive, and perload and skew are unused" | halftoneWord mergeWord mergeFnwith destWord | <inline: false> + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned int, unsigned int)'> + <var: #halftoneWord type: #'unsigned int'> + <var: #mergeWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'unsigned int (*)(unsigned int, unsigned int)'. - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'sqInt (*)(sqInt, sqInt)'. mergeFnwith. "null ref for compiler" 1 to: bbH do: "here is the vertical loop" [ :i | noHalftone ifTrue: [halftoneWord := AllOnes] ifFalse: [halftoneWord := self halftoneAt: dy+i-1]. "Note: the horizontal loop has been expanded into three parts for speed:" "This first section requires masking of the destination store..." destMask := mask1. destWord := self dstLongAt: destIndex. mergeWord := self mergeFn: halftoneWord with: destWord. destWord := (destMask bitAnd: mergeWord) bitOr: (destWord bitAnd: destMask bitInvert32). self dstLongAt: destIndex put: destWord. self incDestIndex: 4. "This central horizontal loop requires no store masking" destMask := AllOnes. combinationRule = 3 ifTrue: ["Special inner loop for STORE" destWord := halftoneWord. 2 to: nWords-1 do:[ :word | self dstLongAt: destIndex put: destWord. self incDestIndex: 4]. ] ifFalse:[ "Normal inner loop does merge" 2 to: nWords-1 do:[ :word | "Normal inner loop does merge" destWord := self dstLongAt: destIndex. mergeWord := self mergeFn: halftoneWord with: destWord. self dstLongAt: destIndex put: mergeWord. self incDestIndex: 4]. ]. "This last section, if used, requires masking of the destination store..." nWords > 1 ifTrue: [destMask := mask2. destWord := self dstLongAt: destIndex. mergeWord := self mergeFn: halftoneWord with: destWord. destWord := (destMask bitAnd: mergeWord) bitOr: (destWord bitAnd: destMask bitInvert32). self dstLongAt: destIndex put: destWord. self incDestIndex: 4]. self incDestIndex: destDelta]! Item was changed: ----- Method: BitBltSimulation>>copyLoopPixMap (in category 'inner loop') ----- copyLoopPixMap "This version of the inner loop maps source pixels to a destination form with different depth. Because it is already unweildy, the loop is not unrolled as in the other versions. Preload, skew and skewMask are all overlooked, since pickSourcePixels delivers its destination word already properly aligned. Note that pickSourcePixels could be copied in-line at the top of the horizontal loop, and some of its inits moved out of the loop." "ar 12/7/1999: The loop has been rewritten to use only one pickSourcePixels call. The idea is that the call itself could be inlined. If we decide not to inline pickSourcePixels we could optimize the loop instead." | skewWord halftoneWord mergeWord scrStartBits nSourceIncs startBits endBits sourcePixMask destPixMask mergeFnwith nPix srcShift dstShift destWord words srcShiftInc dstShiftInc dstShiftLeft mapperFlags | <inline: false> + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned int, unsigned int)'> + <var: #skewWord type: #'unsigned int'> + <var: #halftoneWord type: #'unsigned int'> + <var: #mergeWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> + <var: #sourcePixMask type: #'unsigned int'> + <var: #destPixMask type: #'unsigned int'> + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'unsigned int (*)(unsigned int, unsigned int)'. - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'sqInt (*)(sqInt, sqInt)'. mergeFnwith. "null ref for compiler" "Additional inits peculiar to unequal source and dest pix size..." sourcePPW := 32//sourceDepth. sourcePixMask := maskTable at: sourceDepth. destPixMask := maskTable at: destDepth. mapperFlags := cmFlags bitAnd: ColorMapNewStyle bitInvert32. sourceIndex := sourceBits + (sy * sourcePitch) + ((sx // sourcePPW) *4). scrStartBits := sourcePPW - (sx bitAnd: sourcePPW-1). bbW < scrStartBits ifTrue: [nSourceIncs := 0] ifFalse: [nSourceIncs := (bbW - scrStartBits)//sourcePPW + 1]. sourceDelta := sourcePitch - (nSourceIncs * 4). "Note following two items were already calculated in destmask setup!!" startBits := destPPW - (dx bitAnd: destPPW-1). endBits := ((dx + bbW - 1) bitAnd: destPPW-1) + 1. bbW < startBits ifTrue:[startBits := bbW]. "Precomputed shifts for pickSourcePixels" srcShift := ((sx bitAnd: sourcePPW - 1) * sourceDepth). dstShift := ((dx bitAnd: destPPW - 1) * destDepth). srcShiftInc := sourceDepth. dstShiftInc := destDepth. dstShiftLeft := 0. sourceMSB ifTrue:[ srcShift := 32 - sourceDepth - srcShift. srcShiftInc := 0 - srcShiftInc]. destMSB ifTrue:[ dstShift := 32 - destDepth - dstShift. dstShiftInc := 0 - dstShiftInc. dstShiftLeft := 32 - destDepth]. 1 to: bbH do: "here is the vertical loop" [ :i | "*** is it possible at all that noHalftone == false? ***" noHalftone ifTrue:[halftoneWord := AllOnes] ifFalse: [halftoneWord := self halftoneAt: dy+i-1]. "setup first load" srcBitShift := srcShift. dstBitShift := dstShift. destMask := mask1. nPix := startBits. "Here is the horizontal loop..." words := nWords. ["pick up the word" skewWord := self pickSourcePixels: nPix flags: mapperFlags srcMask: sourcePixMask destMask: destPixMask srcShiftInc: srcShiftInc dstShiftInc: dstShiftInc. "align next word to leftmost pixel" dstBitShift := dstShiftLeft. destMask = AllOnes ifTrue:["avoid read-modify-write" mergeWord := self mergeFn: (skewWord bitAnd: halftoneWord) with: (self dstLongAt: destIndex). self dstLongAt: destIndex put: (destMask bitAnd: mergeWord). ] ifFalse:[ "General version using dest masking" destWord := self dstLongAt: destIndex. mergeWord := self mergeFn: (skewWord bitAnd: halftoneWord) with: (destWord bitAnd: destMask). destWord := (destMask bitAnd: mergeWord) bitOr: (destWord bitAnd: destMask bitInvert32). self dstLongAt: destIndex put: destWord. ]. self incDestIndex: 4. words = 2 "e.g., is the next word the last word?" ifTrue:["set mask for last word in this row" destMask := mask2. nPix := endBits] ifFalse:["use fullword mask for inner loop" destMask := AllOnes. nPix := destPPW]. (words := words - 1) = 0] whileFalse. "--- end of inner loop ---" self incSrcIndex: sourceDelta. self incDestIndex: destDelta] ! Item was changed: ----- Method: BitBltSimulation>>destinationWord:with: (in category 'combination rules') ----- destinationWord: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^destinationWord! Item was changed: ----- Method: BitBltSimulation>>dither32To16:threshold: (in category 'pixel mapping') ----- dither32To16: srcWord threshold: ditherValue "Dither the given 32bit word to 16 bit. Ignore alpha." | addThreshold | <inline: true> "You bet" + <returnTypeC: 'unsigned int'> + <var: #srcWord type: 'unsigned int'> addThreshold := ditherValue bitShift: 8. ^((dither8Lookup at: (addThreshold+((srcWord bitShift: -16) bitAnd: 255))) bitShift: 10) + ((dither8Lookup at: (addThreshold+((srcWord bitShift: -8) bitAnd: 255))) bitShift: 5) + (dither8Lookup at: (addThreshold+(srcWord bitAnd: 255))). ! Item was changed: ----- Method: BitBltSimulation>>expensiveDither32To16:threshold: (in category 'pixel mapping') ----- expensiveDither32To16: srcWord threshold: ditherValue "Dither the given 32bit word to 16 bit. Ignore alpha." | pv threshold value out | <inline: true> "You bet" + <returnTypeC: 'unsigned int'> + <var: #srcWord type: 'unsigned int'> pv := srcWord bitAnd: 255. threshold := ditherThresholds16 at: (pv bitAnd: 7). value := ditherValues16 at: (pv bitShift: -3). ditherValue < threshold ifTrue:[out := value + 1] ifFalse:[out := value]. pv := (srcWord bitShift: -8) bitAnd: 255. threshold := ditherThresholds16 at: (pv bitAnd: 7). value := ditherValues16 at: (pv bitShift: -3). ditherValue < threshold ifTrue:[out := out bitOr: (value+1 bitShift:5)] ifFalse:[out := out bitOr: (value bitShift: 5)]. pv := (srcWord bitShift: -16) bitAnd: 255. threshold := ditherThresholds16 at: (pv bitAnd: 7). value := ditherValues16 at: (pv bitShift: -3). ditherValue < threshold ifTrue:[out := out bitOr: (value+1 bitShift:10)] ifFalse:[out := out bitOr: (value bitShift: 10)]. ^out! Item was changed: ----- Method: BitBltSimulation>>fixAlpha:with: (in category 'combination rules') ----- fixAlpha: sourceWord with: destinationWord "For any non-zero pixel value in destinationWord with zero alpha channel take the alpha from sourceWord and fill it in. Intended for fixing alpha channels left at zero during 16->32 bpp conversions." + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> destDepth = 32 ifFalse:[^destinationWord]. "no-op for non 32bpp" destinationWord = 0 ifTrue:[^0]. (destinationWord bitAnd: 16rFF000000) = 0 ifFalse:[^destinationWord]. ^destinationWord bitOr: (sourceWord bitAnd: 16rFF000000) ! Item was changed: ----- Method: BitBltSimulation>>merge:with: (in category 'combination rules') ----- merge: sourceWord with: destinationWord | mergeFnwith | "Sender warpLoop is too big to include this in-line" + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned int, unsigned int)'> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'unsigned int (*)(unsigned int, unsigned int)'. - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'sqInt (*)(sqInt, sqInt)'. mergeFnwith. "null ref for compiler" ^ self mergeFn: sourceWord with: destinationWord! Item was changed: ----- Method: BitBltSimulation>>partitionedAND:to:nBits:nPartitions: (in category 'combination rules') ----- partitionedAND: word1 to: word2 nBits: nBits nPartitions: nParts "AND word1 to word2 as nParts partitions of nBits each. Any field of word1 not all-ones is treated as all-zeroes. Used for erasing, eg, brush shapes prior to ORing in a color" | mask result | + <returnTypeC: 'unsigned int'> + <var: #word1 type: 'unsigned int'> + <var: #word2 type: 'unsigned int'> + <var: #result type: 'unsigned int'> mask := maskTable at: nBits. "partition mask starts at the right" result := 0. 1 to: nParts do: [:i | (word1 bitAnd: mask) = mask ifTrue: [result := result bitOr: (word2 bitAnd: mask)]. mask := mask << nBits "slide left to next partition"]. ^ result ! Item was changed: ----- Method: BitBltSimulation>>partitionedAdd:to:nBits:componentMask:carryOverflowMask: (in category 'combination rules') ----- partitionedAdd: word1 to: word2 nBits: nBits componentMask: componentMask carryOverflowMask: carryOverflowMask "Add word1 to word2 as nParts partitions of nBits each. This is useful for packed pixels, or packed colors" | carryOverflow sum w1 w2 | "Use unsigned int everywhere because it has a well known arithmetic model without undefined behavior w.r.t. overflow and shifts" + <returnTypeC: 'unsigned int'> <var: #word1 type: 'unsigned int'> <var: #word2 type: 'unsigned int'> <var: #w1 type: 'unsigned int'> <var: #w2 type: 'unsigned int'> <var: #componentMask type: 'unsigned int'> <var: #carryOverflowMask type: 'unsigned int'> <var: #carryOverflow type: 'unsigned int'> <var: #sum type: 'unsigned int'> w1 := word1 bitAnd: carryOverflowMask. "mask to remove high bit of each component" w2 := word2 bitAnd: carryOverflowMask. sum := (word1 bitXor: w1)+(word2 bitXor: w2). "sum without high bit to avoid overflowing over next component" carryOverflow := (w1 bitAnd: w2) bitOr: ((w1 bitOr: w2) bitAnd: sum). "detect overflow condition for saturating" ^((sum bitXor: w1)bitXor:w2) "sum high bit without overflow" bitOr: carryOverflow>>(nBits-1) * componentMask "saturate in case of overflow"! Item was changed: ----- Method: BitBltSimulation>>partitionedMax:with:nBits:nPartitions: (in category 'combination rules') ----- partitionedMax: word1 with: word2 nBits: nBits nPartitions: nParts "Max word1 to word2 as nParts partitions of nBits each" | mask result | "In C, most arithmetic operations answer the same bit pattern regardless of the operands being signed or unsigned ints (this is due to the way 2's complement numbers work). However, comparisions might fail. Add the proper declaration of words as unsigned int in those cases where comparisions are done (jmv)" + <returnTypeC: 'unsigned int'> <var: #word1 type: 'unsigned int'> <var: #word2 type: 'unsigned int'> <var: #mask type: 'unsigned int'> <var: #result type: 'unsigned int'> mask := maskTable at: nBits. "partition mask starts at the right" result := 0. 1 to: nParts do: [:i | result := result bitOr: ((word2 bitAnd: mask) max: (word1 bitAnd: mask)). mask := mask << nBits "slide left to next partition"]. ^ result ! Item was changed: ----- Method: BitBltSimulation>>partitionedMin:with:nBits:nPartitions: (in category 'combination rules') ----- partitionedMin: word1 with: word2 nBits: nBits nPartitions: nParts "Min word1 to word2 as nParts partitions of nBits each" | mask result | "In C, most arithmetic operations answer the same bit pattern regardless of the operands being signed or unsigned ints (this is due to the way 2's complement numbers work). However, comparisions might fail. Add the proper declaration of words as unsigned int in those cases where comparisions are done (jmv)" + <returnTypeC: 'unsigned int'> <var: #word1 type: 'unsigned int'> <var: #word2 type: 'unsigned int'> <var: #mask type: 'unsigned int'> <var: #result type: 'unsigned int'> mask := maskTable at: nBits. "partition mask starts at the right" result := 0. 1 to: nParts do: [:i | result := result bitOr: ((word2 bitAnd: mask) min: (word1 bitAnd: mask)). mask := mask << nBits "slide left to next partition"]. ^ result ! Item was changed: ----- Method: BitBltSimulation>>partitionedMul:with:nBits:nPartitions: (in category 'combination rules') ----- partitionedMul: word1 with: word2 nBits: nBits nPartitions: nParts "Multiply word1 with word2 as nParts partitions of nBits each. This is useful for packed pixels, or packed colors. Bug in loop version when non-white background" | sMask product result dMask | "In C, integer multiplication might answer a wrong value if the unsigned values are declared as signed. This problem does not affect this method, because the most significant bit (i.e. the sign bit) will always be zero (jmv)" + <returnTypeC: 'unsigned int'> + <var: #word1 type: 'unsigned int'> + <var: #word2 type: 'unsigned int'> + <var: #sMask type: 'unsigned int'> + <var: #dMask type: 'unsigned int'> + <var: #result type: 'unsigned int'> + <var: #product type: 'unsigned int'> sMask := maskTable at: nBits. "partition mask starts at the right" dMask := sMask << nBits. result := (((word1 bitAnd: sMask)+1) * ((word2 bitAnd: sMask)+1) - 1 bitAnd: dMask) >> nBits. "optimized first step" nParts = 1 ifTrue: [ ^result ]. product := (((word1>>nBits bitAnd: sMask)+1) * ((word2>>nBits bitAnd: sMask)+1) - 1 bitAnd: dMask). result := result bitOr: product. nParts = 2 ifTrue: [ ^result ]. product := (((word1>>(2*nBits) bitAnd: sMask)+1) * ((word2>>(2*nBits) bitAnd: sMask)+1) - 1 bitAnd: dMask). result := result bitOr: product << nBits. nParts = 3 ifTrue: [ ^result ]. product := (((word1>>(3*nBits) bitAnd: sMask)+1) * ((word2>>(3*nBits) bitAnd: sMask)+1) - 1 bitAnd: dMask). result := result bitOr: product << (2*nBits). ^ result " | sMask product result dMask | sMask := maskTable at: nBits. 'partition mask starts at the right' dMask := sMask << nBits. result := (((word1 bitAnd: sMask)+1) * ((word2 bitAnd: sMask)+1) - 1 bitAnd: dMask) >> nBits. 'optimized first step' nBits to: nBits * (nParts-1) by: nBits do: [:ofs | product := (((word1>>ofs bitAnd: sMask)+1) * ((word2>>ofs bitAnd: sMask)+1) - 1 bitAnd: dMask). result := result bitOr: (product bitAnd: dMask) << (ofs-nBits)]. ^ result"! Item was changed: ----- Method: BitBltSimulation>>partitionedRgbComponentAlpha:dest:nBits:nPartitions: (in category 'combination rules') ----- partitionedRgbComponentAlpha: sourceWord dest: destWord nBits: nBits nPartitions: nParts | mask result p1 p2 v | + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destWord type: 'unsigned int'> + <var: #p1 type: 'unsigned int'> + <var: #p2 type: 'unsigned int'> + <var: #mask type: 'unsigned int'> + <var: #result type: 'unsigned int'> mask := maskTable at: nBits. "partition mask starts at the right" result := 0. 1 to: nParts do: [:i | p1 := (sourceWord bitAnd: mask) >> ((i - 1)*nBits). p2 := (destWord bitAnd: mask) >> ((i - 1)*nBits). nBits = 32 ifFalse:[ nBits = 16 ifTrue:[ p1 := (self rgbMap16To32: p1) bitOr: 16rFF000000. p2 := (self rgbMap16To32: p2) bitOr: 16rFF000000] ifFalse:[ p1 := (self rgbMap: p1 from: nBits to: 32) bitOr: 16rFF000000. p2 := (self rgbMap: p2 from: nBits to: 32) bitOr: 16rFF000000.]]. v := self rgbComponentAlpha32: p1 with: p2. nBits = 32 ifFalse:[ v := self rgbMap: v from: 32 to: nBits]. result := result bitOr: (v << ((i - 1)*nBits)). mask := mask << nBits "slide left to next partition"]. ^ result ! Item was changed: ----- Method: BitBltSimulation>>partitionedSub:from:nBits:nPartitions: (in category 'combination rules') ----- partitionedSub: word1 from: word2 nBits: nBits nPartitions: nParts "Subtract word1 from word2 as nParts partitions of nBits each. This is useful for packed pixels, or packed colors" | mask result p1 p2 | "In C, most arithmetic operations answer the same bit pattern regardless of the operands being signed or unsigned ints (this is due to the way 2's complement numbers work). However, comparisions might fail. Add the proper declaration of words as unsigned int in those cases where comparisions are done (jmv)" + <returnTypeC: 'unsigned int'> <var: #word1 type: 'unsigned int'> <var: #word2 type: 'unsigned int'> <var: #p1 type: 'unsigned int'> <var: #p2 type: 'unsigned int'> <var: #mask type: 'unsigned int'> <var: #result type: 'unsigned int'> mask := maskTable at: nBits. "partition mask starts at the right" result := 0. 1 to: nParts do: [:i | p1 := word1 bitAnd: mask. p2 := word2 bitAnd: mask. p1 < p2 "result is really abs value of thedifference" ifTrue: [result := result bitOr: p2 - p1] ifFalse: [result := result bitOr: p1 - p2]. mask := mask << nBits "slide left to next partition"]. ^ result ! Item was changed: ----- Method: BitBltSimulation>>pickSourcePixels:flags:srcMask:destMask:srcShiftInc:dstShiftInc: (in category 'combination rules') ----- pickSourcePixels: nPixels flags: mapperFlags srcMask: srcMask destMask: dstMask srcShiftInc: srcShiftInc dstShiftInc: dstShiftInc "Pick nPix pixels starting at srcBitIndex from the source, map by the color map, and justify them according to dstBitIndex in the resulting destWord." | sourceWord destWord sourcePix destPix srcShift dstShift nPix | <inline: true> "oh please" + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> sourceWord := self srcLongAt: sourceIndex. destWord := 0. srcShift := srcBitShift. "Hint: Keep in register" dstShift := dstBitShift. "Hint: Keep in register" nPix := nPixels. "always > 0 so we can use do { } while(--nPix);" (mapperFlags = (ColorMapPresent bitOr: ColorMapIndexedPart)) ifTrue:[ "a little optimization for (pretty crucial) blits using indexed lookups only" [ "grab, colormap and mix in pixel" sourcePix := sourceWord >> srcShift bitAnd: srcMask. destPix := cmLookupTable at: (sourcePix bitAnd: cmMask). destWord := destWord bitOr: (destPix bitAnd: dstMask) << dstShift. "adjust dest pix index" dstShift := dstShift + dstShiftInc. "adjust source pix index" ((srcShift := srcShift + srcShiftInc) bitAnd: 16rFFFFFFE0) = 0 ifFalse:[ sourceMSB ifTrue:[srcShift := srcShift + 32] ifFalse:[srcShift := srcShift - 32]. sourceWord := self srcLongAt: (self incSrcIndex: 4)]. (nPix := nPix - 1) = 0] whileFalse. ] ifFalse:[ [ "grab, colormap and mix in pixel" sourcePix := sourceWord >> srcShift bitAnd: srcMask. destPix := self mapPixel: sourcePix flags: mapperFlags. destWord := destWord bitOr: (destPix bitAnd: dstMask) << dstShift. "adjust dest pix index" dstShift := dstShift + dstShiftInc. "adjust source pix index" ((srcShift := srcShift + srcShiftInc) bitAnd: 16rFFFFFFE0) = 0 ifFalse:[ sourceMSB ifTrue:[srcShift := srcShift + 32] ifFalse:[srcShift := srcShift - 32]. sourceWord := self srcLongAt: (self incSrcIndex: 4)]. (nPix := nPix - 1) = 0] whileFalse. ]. srcBitShift := srcShift. "Store back" ^destWord ! Item was changed: ----- Method: BitBltSimulation>>pickWarpPixelAtX:y: (in category 'pixel mapping') ----- pickWarpPixelAtX: xx y: yy "Pick a single pixel from the source for WarpBlt. Note: This method is crucial for WarpBlt speed w/o smoothing and still relatively important when smoothing is used." | x y srcIndex sourceWord sourcePix | <inline: true> "*please*" + <returnTypeC: #'unsigned int'> + <var: #sourceWord type: #'unsigned int'> "note: it would be much faster if we could just avoid these stupid tests for being inside sourceForm." (xx < 0 or:[yy < 0 or:[ (x := xx >> BinaryPoint) >= sourceWidth or:[ (y := yy >> BinaryPoint) >= sourceHeight]]]) ifTrue:[^0]. "out of bounds" "Fetch source word. Note: We should really update srcIndex with sx and sy so that we don't have to do the computation below. We might even be able to simplify the out of bounds test from above." srcIndex := sourceBits + (y * sourcePitch) + (x >> warpAlignShift * 4). sourceWord := self srcLongAt: srcIndex. "Extract pixel from word" srcBitShift := warpBitShiftTable at: (x bitAnd: warpAlignMask). sourcePix := sourceWord >> srcBitShift bitAnd: warpSrcMask. ^sourcePix! Item was changed: ----- Method: BitBltSimulation>>pixClear:with: (in category 'combination rules') ----- pixClear: sourceWord with: destinationWord "Clear all pixels in destinationWord for which the pixels of sourceWord have the same values. Used to clear areas of some constant color to zero." | mask result nBits pv | <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> + <var: #mask type: 'unsigned int'> + <var: #pv type: 'unsigned int'> + <var: #result type: 'unsigned int'> destDepth = 32 ifTrue:[ sourceWord = destinationWord ifTrue:[^0] ifFalse:[^destinationWord]. ]. nBits := destDepth. mask := maskTable at: nBits. "partition mask starts at the right" result := 0. 1 to: destPPW do:[:i | pv := destinationWord bitAnd: mask. (sourceWord bitAnd: mask) = pv ifTrue:[pv := 0]. result := result bitOr: pv. mask := mask << nBits "slide left to next partition"]. ^ result! Item was changed: ----- Method: BitBltSimulation>>pixMask:with: (in category 'combination rules') ----- pixMask: sourceWord with: destinationWord <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^ self partitionedAND: sourceWord bitInvert32 to: destinationWord nBits: destDepth nPartitions: destPPW! Item was changed: ----- Method: BitBltSimulation>>pixPaint:with: (in category 'combination rules') ----- pixPaint: sourceWord with: destinationWord <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> sourceWord = 0 ifTrue: [^ destinationWord]. ^ sourceWord bitOr: (self partitionedAND: sourceWord bitInvert32 to: destinationWord nBits: destDepth nPartitions: destPPW)! Item was changed: ----- Method: BitBltSimulation>>pixSwap:with: (in category 'combination rules') ----- pixSwap: sourceWord with: destWord "Swap the pixels in destWord" | result shift lowMask highMask | <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destWord type: 'unsigned int'> + <var: #lowMask type: 'unsigned int'> + <var: #highMask type: 'unsigned int'> + <var: #result type: 'unsigned int'> destPPW = 1 ifTrue:[^destWord]. "a single pixel per word" result := 0. lowMask := (1 << destDepth) - 1. "mask low pixel" highMask := lowMask << (destPPW-1 * destDepth). "mask high pixel" shift := 32 - destDepth. result := result bitOr: ( (destWord bitAnd: lowMask) << shift bitOr: (destWord bitAnd: highMask) >> shift). destPPW <= 2 ifTrue:[^result]. 2 to: destPPW // 2 do:[:i| lowMask := lowMask << destDepth. highMask := highMask >> destDepth. shift := shift - (destDepth * 2). result := result bitOr: ( (destWord bitAnd: lowMask) << shift bitOr: (destWord bitAnd: highMask) >> shift)]. ^result! Item was changed: ----- Method: BitBltSimulation>>rgbAdd:with: (in category 'combination rules') ----- rgbAdd: sourceWord with: destinationWord <inline: false> + <returnTypeC: 'unsigned int'> <var: #sourceWord type: 'unsigned int'> <var: #destinationWord type: 'unsigned int'> <var: #carryOverflowMask type: 'unsigned int'> <var: #componentMask type: 'unsigned int'> | componentMask carryOverflowMask | destDepth < 16 ifTrue: ["Add each pixel separately" componentMask := 1<<destDepth-1. carryOverflowMask := 16rFFFFFFFF//componentMask<<(destDepth-1). ^ self partitionedAdd: sourceWord to: destinationWord nBits: destDepth componentMask: componentMask carryOverflowMask: carryOverflowMask]. destDepth = 16 ifTrue: ["Add RGB components of each pixel separately" componentMask := 16r1F. carryOverflowMask := 16r42104210. ^ (self partitionedAdd: (sourceWord bitAnd: 16r7FFF7FFF) to: (destinationWord bitAnd: 16r7FFF7FFF) "make sure that the unused bit is at 0" nBits: 5 componentMask: componentMask carryOverflowMask: carryOverflowMask)] ifFalse: ["Add RGBA components of the pixel separately" componentMask := 16rFF. carryOverflowMask := 16r80808080. ^ self partitionedAdd: sourceWord to: destinationWord nBits: 8 componentMask: componentMask carryOverflowMask: carryOverflowMask]! Item was changed: ----- Method: BitBltSimulation>>rgbComponentAlpha16 (in category 'combination rules') ----- rgbComponentAlpha16 "This version assumes combinationRule = 41 sourcePixSize = 32 destPixSize = 16 sourceForm ~= destForm. " <inline: false> "This particular method should be optimized in itself" + <var: #sourceWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> + <var: #dstMask type: #'unsigned int'> | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY srcY dstY dstMask srcShift ditherBase ditherIndex ditherThreshold | deltaY := bbH + 1. "So we can pre-decrement" srcY := sy. dstY := dy. srcShift := (dx bitAnd: 1) * 16. destMSB ifTrue:[srcShift := 16 - srcShift]. mask1 := 16rFFFF << (16 - srcShift). "This is the outer loop" [(deltaY := deltaY - 1) ~= 0] whileTrue:[ srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). dstIndex := destBits + (dstY * destPitch) + (dx // 2 * 4). ditherBase := (dstY bitAnd: 3) * 4. ditherIndex := (sx bitAnd: 3) - 1. "For pre-increment" deltaX := bbW + 1. "So we can pre-decrement" dstMask := mask1. dstMask = 16rFFFF ifTrue:[srcShift := 16] ifFalse:[srcShift := 0]. "This is the inner loop" [(deltaX := deltaX - 1) ~= 0] whileTrue:[ ditherThreshold := ditherMatrix4x4 at: ditherBase + (ditherIndex := ditherIndex + 1 bitAnd: 3). sourceWord := self srcLongAt: srcIndex. srcAlpha := sourceWord bitAnd: 16rFFFFFF. srcAlpha = 0 ifFalse:[ "0 < srcAlpha" "If we have to mix colors then just copy a single word" destWord := self dstLongAt: dstIndex. destWord := destWord bitAnd: dstMask bitInvert32. destWord := destWord >> srcShift. "Expand from 16 to 32 bit by adding zero bits" destWord := (((destWord bitAnd: 16r7C00) bitShift: 9) bitOr: ((destWord bitAnd: 16r3E0) bitShift: 6)) bitOr: (((destWord bitAnd: 16r1F) bitShift: 3) bitOr: 16rFF000000). "Mix colors" sourceWord := self rgbComponentAlpha32: sourceWord with: destWord. "And dither" sourceWord := self dither32To16: sourceWord threshold: ditherThreshold. sourceWord = 0 ifTrue:[sourceWord := 1 << srcShift] ifFalse:[sourceWord := sourceWord << srcShift]. "Store back" self dstLongAt: dstIndex put: sourceWord mask: dstMask. ]. srcIndex := srcIndex + 4. destMSB ifTrue:[srcShift = 0 ifTrue:[dstIndex := dstIndex + 4]] ifFalse:[srcShift = 0 ifFalse:[dstIndex := dstIndex + 4]]. srcShift := srcShift bitXor: 16. "Toggle between 0 and 16" dstMask := dstMask bitInvert32. "Mask other half word" ]. srcY := srcY + 1. dstY := dstY + 1. ]. ! Item was changed: ----- Method: BitBltSimulation>>rgbComponentAlpha32 (in category 'combination rules') ----- rgbComponentAlpha32 "This version assumes combinationRule = 41 sourcePixSize = destPixSize = 32 sourceForm ~= destForm. Note: The inner loop has been optimized for dealing with the special case of aR = aG = aB = 0 " | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY srcY dstY | <inline: false> "This particular method should be optimized in itself" "Give the compile a couple of hints" - <var: #sourceWord type: 'register long'> <var: #deltaX type: 'register long'> + <var: #sourceWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> "The following should be declared as pointers so the compiler will notice that they're used for accessing memory locations (good to know on an Intel architecture) but then the increments would be different between ST code and C code so must hope the compiler notices what happens (MS Visual C does)" <var: #srcIndex type: 'register long'> <var: #dstIndex type: 'register long'> deltaY := bbH + 1. "So we can pre-decrement" srcY := sy. dstY := dy. "This is the outer loop" [(deltaY := deltaY - 1) ~= 0] whileTrue:[ srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). dstIndex := destBits + (dstY * destPitch) + (dx * 4). deltaX := bbW + 1. "So we can pre-decrement" "This is the inner loop" [(deltaX := deltaX - 1) ~= 0] whileTrue:[ sourceWord := self srcLongAt: srcIndex. srcAlpha := sourceWord bitAnd:16rFFFFFF. srcAlpha = 0 ifTrue:[ srcIndex := srcIndex + 4. dstIndex := dstIndex + 4. "Now skip as many words as possible," [(deltaX := deltaX - 1) ~= 0 and:[ ((sourceWord := self srcLongAt: srcIndex) bitAnd:16rFFFFFF) = 0]] whileTrue:[ srcIndex := srcIndex + 4. dstIndex := dstIndex + 4. ]. "Adjust deltaX" deltaX := deltaX + 1. ] ifFalse:[ "0 < srcAlpha" "If we have to mix colors then just copy a single word" destWord := self dstLongAt: dstIndex. destWord := self rgbComponentAlpha32: sourceWord with: destWord. self dstLongAt: dstIndex put: destWord. srcIndex := srcIndex + 4. dstIndex := dstIndex + 4. ]. ]. srcY := srcY + 1. dstY := dstY + 1. ].! Item was changed: ----- Method: BitBltSimulation>>rgbComponentAlpha8 (in category 'combination rules') ----- rgbComponentAlpha8 "This version assumes combinationRule = 41 sourcePixSize = 32 destPixSize = 8 sourceForm ~= destForm. Note: This is not real blending since we don't have the source colors available. " | srcIndex dstIndex sourceWord srcAlpha destWord deltaX deltaY srcY dstY dstMask srcShift adjust mappingTable mapperFlags | <inline: false> "This particular method should be optimized in itself" <var: #mappingTable declareC:'unsigned int *mappingTable'> + <var: #sourceWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> + <var: #dstMask type: #'unsigned int'> mappingTable := self default8To32Table. mapperFlags := cmFlags bitAnd: ColorMapNewStyle bitInvert32. deltaY := bbH + 1. "So we can pre-decrement" srcY := sy. dstY := dy. mask1 := ((dx bitAnd: 3) * 8). destMSB ifTrue:[mask1 := 24 - mask1]. mask2 := AllOnes bitXor:(16rFF << mask1). (dx bitAnd: 1) = 0 ifTrue:[adjust := 0] ifFalse:[adjust := 16r1F1F1F1F]. (dy bitAnd: 1) = 0 ifTrue:[adjust := adjust bitXor: 16r1F1F1F1F]. "This is the outer loop" [(deltaY := deltaY - 1) ~= 0] whileTrue:[ adjust := adjust bitXor: 16r1F1F1F1F. srcIndex := sourceBits + (srcY * sourcePitch) + (sx * 4). dstIndex := destBits + (dstY * destPitch) + (dx // 4 * 4). deltaX := bbW + 1. "So we can pre-decrement" srcShift := mask1. dstMask := mask2. "This is the inner loop" [(deltaX := deltaX - 1) ~= 0] whileTrue:[ sourceWord := ((self srcLongAt: srcIndex) bitAnd: (adjust bitInvert32)) + adjust. srcAlpha := sourceWord bitAnd: 16rFFFFFF. "set srcAlpha to the average of the 3 separate aR,Ag,AB values" srcAlpha := ((srcAlpha >> 16) + (srcAlpha >> 8 bitAnd: 16rFF) + (srcAlpha bitAnd: 16rFF)) // 3. srcAlpha > 31 ifTrue:["Everything below 31 is transparent" srcAlpha > 224 ifTrue: ["treat everything above 224 as opaque" sourceWord := 16rFFFFFFFF]. destWord := self dstLongAt: dstIndex. destWord := destWord bitAnd: dstMask bitInvert32. destWord := destWord >> srcShift. destWord := mappingTable at: destWord. sourceWord := self rgbComponentAlpha32: sourceWord with: destWord. sourceWord := self mapPixel: sourceWord flags: mapperFlags. sourceWord := sourceWord << srcShift. "Store back" self dstLongAt: dstIndex put: sourceWord mask: dstMask. ]. srcIndex := srcIndex + 4. destMSB ifTrue:[ srcShift = 0 ifTrue:[dstIndex := dstIndex + 4. srcShift := 24. dstMask := 16r00FFFFFF] ifFalse:[srcShift := srcShift - 8. dstMask := (dstMask >> 8) bitOr: 16rFF000000]. ] ifFalse:[ srcShift = 32 ifTrue:[dstIndex := dstIndex + 4. srcShift := 0. dstMask := 16rFFFFFF00] ifFalse:[srcShift := srcShift + 8. dstMask := dstMask << 8 bitOr: 255]. ]. adjust := adjust bitXor: 16r1F1F1F1F. ]. srcY := srcY + 1. dstY := dstY + 1. ]. ! Item was changed: ----- Method: BitBltSimulation>>rgbDiff:with: (in category 'combination rules') ----- rgbDiff: sourceWord with: destinationWord "Subract the pixels in the source and destination, color by color, and return the sum of the absolute value of all the differences. For non-rgb, return the number of differing pixels." | pixMask destShifted sourceShifted destPixVal bitsPerColor rgbMask sourcePixVal diff maskShifted | <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> + <var: #sourceShifted type: 'unsigned int'> + <var: #destShifted type: 'unsigned int'> + <var: #maskShifted type: 'unsigned int'> + <var: #pixMask type: 'unsigned int'> + <var: #rgbMask type: 'unsigned int'> pixMask := maskTable at: destDepth. destDepth = 16 ifTrue: [bitsPerColor := 5. rgbMask := 16r1F] ifFalse: [bitsPerColor := 8. rgbMask := 16rFF]. maskShifted := destMask. destShifted := destinationWord. sourceShifted := sourceWord. 1 to: destPPW do: [:i | (maskShifted bitAnd: pixMask) > 0 ifTrue: ["Only tally pixels within the destination rectangle" destPixVal := destShifted bitAnd: pixMask. sourcePixVal := sourceShifted bitAnd: pixMask. destDepth < 16 ifTrue: [sourcePixVal = destPixVal ifTrue: [diff := 0] ifFalse: [diff := 1]] ifFalse: [diff := (self partitionedSub: sourcePixVal from: destPixVal nBits: bitsPerColor nPartitions: 3). diff := (diff bitAnd: rgbMask) + (diff>>bitsPerColor bitAnd: rgbMask) + ((diff>>bitsPerColor)>>bitsPerColor bitAnd: rgbMask)]. bitCount := bitCount + diff]. maskShifted := maskShifted >> destDepth. sourceShifted := sourceShifted >> destDepth. destShifted := destShifted >> destDepth]. ^ destinationWord "For no effect on dest" ! Item was changed: ----- Method: BitBltSimulation>>rgbMax:with: (in category 'combination rules') ----- rgbMax: sourceWord with: destinationWord <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> destDepth < 16 ifTrue: ["Max each pixel separately" ^ self partitionedMax: sourceWord with: destinationWord nBits: destDepth nPartitions: destPPW]. destDepth = 16 ifTrue: ["Max RGB components of each pixel separately" ^ (self partitionedMax: sourceWord with: destinationWord nBits: 5 nPartitions: 3) + ((self partitionedMax: sourceWord>>16 with: destinationWord>>16 nBits: 5 nPartitions: 3) << 16)] ifFalse: ["Max RGBA components of the pixel separately" ^ self partitionedMax: sourceWord with: destinationWord nBits: 8 nPartitions: 4]! Item was changed: ----- Method: BitBltSimulation>>rgbMin:with: (in category 'combination rules') ----- rgbMin: sourceWord with: destinationWord <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> destDepth < 16 ifTrue: ["Min each pixel separately" ^ self partitionedMin: sourceWord with: destinationWord nBits: destDepth nPartitions: destPPW]. destDepth = 16 ifTrue: ["Min RGB components of each pixel separately" ^ (self partitionedMin: sourceWord with: destinationWord nBits: 5 nPartitions: 3) + ((self partitionedMin: sourceWord>>16 with: destinationWord>>16 nBits: 5 nPartitions: 3) << 16)] ifFalse: ["Min RGBA components of the pixel separately" ^ self partitionedMin: sourceWord with: destinationWord nBits: 8 nPartitions: 4]! Item was changed: ----- Method: BitBltSimulation>>rgbMinInvert:with: (in category 'combination rules') ----- rgbMinInvert: wordToInvert with: destinationWord | sourceWord | <inline: false> + <returnTypeC: 'unsigned int'> + <var: #wordToInvert type: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> sourceWord := wordToInvert bitInvert32. destDepth < 16 ifTrue: ["Min each pixel separately" ^ self partitionedMin: sourceWord with: destinationWord nBits: destDepth nPartitions: destPPW]. destDepth = 16 ifTrue: ["Min RGB components of each pixel separately" ^ (self partitionedMin: sourceWord with: destinationWord nBits: 5 nPartitions: 3) + ((self partitionedMin: sourceWord>>16 with: destinationWord>>16 nBits: 5 nPartitions: 3) << 16)] ifFalse: ["Min RGBA components of the pixel separately" ^ self partitionedMin: sourceWord with: destinationWord nBits: 8 nPartitions: 4]! Item was changed: ----- Method: BitBltSimulation>>rgbMul:with: (in category 'combination rules') ----- rgbMul: sourceWord with: destinationWord <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> destDepth < 16 ifTrue: ["Mul each pixel separately" ^ self partitionedMul: sourceWord with: destinationWord nBits: destDepth nPartitions: destPPW]. destDepth = 16 ifTrue: ["Mul RGB components of each pixel separately" ^ (self partitionedMul: sourceWord with: destinationWord nBits: 5 nPartitions: 3) + ((self partitionedMul: sourceWord>>16 with: destinationWord>>16 nBits: 5 nPartitions: 3) << 16)] ifFalse: ["Mul RGBA components of the pixel separately" ^ self partitionedMul: sourceWord with: destinationWord nBits: 8 nPartitions: 4] " | scanner | Display repaintMorphicDisplay. scanner := DisplayScanner quickPrintOn: Display. MessageTally time: [0 to: 760 by: 4 do: [:y |scanner drawString: 'qwrepoiuasfd=)(/&()=#!!lkjzxv.,mn124+09857907QROIYTOAFDJZXNBNB,M-.,Mqwrepoiuasfd=)(/&()=#!!lkjzxv.,mn124+09857907QROIYTOAFDJZXNBNB,M-.,M1234124356785678' at: 0@y]]. "! Item was changed: ----- Method: BitBltSimulation>>rgbSub:with: (in category 'combination rules') ----- rgbSub: sourceWord with: destinationWord <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> destDepth < 16 ifTrue: ["Sub each pixel separately" ^ self partitionedSub: sourceWord from: destinationWord nBits: destDepth nPartitions: destPPW]. destDepth = 16 ifTrue: ["Sub RGB components of each pixel separately" ^ (self partitionedSub: sourceWord from: destinationWord nBits: 5 nPartitions: 3) + ((self partitionedSub: sourceWord>>16 from: destinationWord>>16 nBits: 5 nPartitions: 3) << 16)] ifFalse: ["Sub RGBA components of the pixel separately" ^ self partitionedSub: sourceWord from: destinationWord nBits: 8 nPartitions: 4]! Item was changed: ----- Method: BitBltSimulation>>sourceWord:with: (in category 'combination rules') ----- sourceWord: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord! Item was changed: ----- Method: BitBltSimulation>>subWord:with: (in category 'combination rules') ----- subWord: sourceWord with: destinationWord + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> ^sourceWord - destinationWord! Item was changed: ----- Method: BitBltSimulation>>tallyIntoMap:with: (in category 'combination rules') ----- tallyIntoMap: sourceWord with: destinationWord "Tally pixels into the color map. Those tallied are exactly those in the destination rectangle. Note that the source should be specified == destination, in order for the proper color map checks to be performed at setup." | mapIndex pixMask destShifted maskShifted pixVal | <inline: false> + <returnTypeC: 'unsigned int'> + <var: #sourceWord type: 'unsigned int'> + <var: #destinationWord type: 'unsigned int'> + <var: #pixMask type: 'unsigned int'> + <var: #destShifted type: 'unsigned int'> + <var: #maskShifted type: 'unsigned int'> (cmFlags bitAnd: (ColorMapPresent bitOr: ColorMapIndexedPart)) = (ColorMapPresent bitOr: ColorMapIndexedPart) ifFalse: [^ destinationWord "no op"]. pixMask := maskTable at: destDepth. destShifted := destinationWord. maskShifted := destMask. 1 to: destPPW do: [:i | (maskShifted bitAnd: pixMask) = 0 ifFalse: ["Only tally pixels within the destination rectangle" pixVal := destShifted bitAnd: pixMask. destDepth < 16 ifTrue: [mapIndex := pixVal] ifFalse: [destDepth = 16 ifTrue: [mapIndex := self rgbMap: pixVal from: 5 to: cmBitsPerColor] ifFalse: [mapIndex := self rgbMap: pixVal from: 8 to: cmBitsPerColor]]. self tallyMapAt: mapIndex put: (self tallyMapAt: mapIndex) + 1]. maskShifted := maskShifted >> destDepth. destShifted := destShifted >> destDepth]. ^ destinationWord "For no effect on dest"! Item was changed: ----- Method: BitBltSimulation>>warpLoop (in category 'inner loop') ----- warpLoop "This version of the inner loop traverses an arbirary quadrilateral source, thus producing a general affine transformation." | skewWord halftoneWord mergeWord startBits deltaP12x deltaP12y deltaP43x deltaP43y pAx pAy pBx pBy xDelta yDelta smoothingCount sourceMapOop nSteps nPix words destWord endBits mergeFnwith dstShiftInc dstShiftLeft mapperFlags | + <inline: false> + <var: #mergeFnwith declareC: 'unsigned int (*mergeFnwith)(unsigned int, unsigned int)'> + <var: #skewWord type: #'unsigned int'> + <var: #halftoneWord type: #'unsigned int'> + <var: #mergeWord type: #'unsigned int'> + <var: #destWord type: #'unsigned int'> + mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'unsigned int (*)(unsigned int, unsigned int)'. - <inline: false> - <var: #mergeFnwith declareC: 'sqInt (*mergeFnwith)(sqInt, sqInt)'> - mergeFnwith := self cCoerce: (opTable at: combinationRule+1) to: 'sqInt (*)(sqInt, sqInt)'. mergeFnwith. "null ref for compiler" (interpreterProxy slotSizeOf: bitBltOop) >= (BBWarpBase+12) ifFalse: [^ interpreterProxy primitiveFail]. nSteps := height-1. nSteps <= 0 ifTrue: [nSteps := 1]. pAx := self fetchIntOrFloat: BBWarpBase ofObject: bitBltOop. words := self fetchIntOrFloat: BBWarpBase+3 ofObject: bitBltOop. deltaP12x := self deltaFrom: pAx to: words nSteps: nSteps. deltaP12x < 0 ifTrue: [pAx := words - (nSteps*deltaP12x)]. pAy := self fetchIntOrFloat: BBWarpBase+1 ofObject: bitBltOop. words := self fetchIntOrFloat: BBWarpBase+4 ofObject: bitBltOop. deltaP12y := self deltaFrom: pAy to: words nSteps: nSteps. deltaP12y < 0 ifTrue: [pAy := words - (nSteps*deltaP12y)]. pBx := self fetchIntOrFloat: BBWarpBase+9 ofObject: bitBltOop. words := self fetchIntOrFloat: BBWarpBase+6 ofObject: bitBltOop. deltaP43x := self deltaFrom: pBx to: words nSteps: nSteps. deltaP43x < 0 ifTrue: [pBx := words - (nSteps*deltaP43x)]. pBy := self fetchIntOrFloat: BBWarpBase+10 ofObject: bitBltOop. words := self fetchIntOrFloat: BBWarpBase+7 ofObject: bitBltOop. deltaP43y := self deltaFrom: pBy to: words nSteps: nSteps. deltaP43y < 0 ifTrue: [pBy := words - (nSteps*deltaP43y)]. interpreterProxy failed ifTrue: [^ false]. "ie if non-integers above" interpreterProxy methodArgumentCount = 2 ifTrue: [smoothingCount := interpreterProxy stackIntegerValue: 1. sourceMapOop := interpreterProxy stackValue: 0. sourceMapOop = interpreterProxy nilObject ifTrue: [sourceDepth < 16 ifTrue: ["color map is required to smooth non-RGB dest" ^ interpreterProxy primitiveFail]] ifFalse: [(interpreterProxy slotSizeOf: sourceMapOop) < (1 << sourceDepth) ifTrue: ["sourceMap must be long enough for sourceDepth" ^ interpreterProxy primitiveFail]. sourceMapOop := self oopForPointer: (interpreterProxy firstIndexableField: sourceMapOop)]] ifFalse: [smoothingCount := 1. sourceMapOop := interpreterProxy nilObject]. nSteps := width-1. nSteps <= 0 ifTrue: [nSteps := 1]. startBits := destPPW - (dx bitAnd: destPPW-1). endBits := ((dx + bbW - 1) bitAnd: destPPW-1) + 1. bbW < startBits ifTrue:[startBits := bbW]. destY < clipY ifTrue:[ "Advance increments if there was clipping in y" pAx := pAx + (clipY - destY * deltaP12x). pAy := pAy + (clipY - destY * deltaP12y). pBx := pBx + (clipY - destY * deltaP43x). pBy := pBy + (clipY - destY * deltaP43y)]. "Setup values for faster pixel fetching." self warpLoopSetup. "Setup color mapping if not provided" (smoothingCount > 1 and:[(cmFlags bitAnd: ColorMapNewStyle) = 0]) ifTrue:[ cmLookupTable == nil ifTrue:[ destDepth = 16 ifTrue:[self setupColorMasksFrom: 8 to: 5]. ] ifFalse:[ self setupColorMasksFrom: 8 to: cmBitsPerColor. ]. ]. mapperFlags := cmFlags bitAnd: ColorMapNewStyle bitInvert32. destMSB ifTrue:[ dstShiftInc := 0 - destDepth. dstShiftLeft := 32 - destDepth] ifFalse:[ dstShiftInc := destDepth. dstShiftLeft := 0]. 1 to: bbH do: [ :i | "here is the vertical loop..." xDelta := self deltaFrom: pAx to: pBx nSteps: nSteps. xDelta >= 0 ifTrue: [sx := pAx] ifFalse: [sx := pBx - (nSteps*xDelta)]. yDelta := self deltaFrom: pAy to: pBy nSteps: nSteps. yDelta >= 0 ifTrue: [sy := pAy] ifFalse: [sy := pBy - (nSteps*yDelta)]. destMSB ifTrue:[dstBitShift := 32 - ((dx bitAnd: destPPW - 1) + 1 * destDepth)] ifFalse:[dstBitShift := (dx bitAnd: destPPW - 1) * destDepth]. (destX < clipX) ifTrue:[ "Advance increments if there was clipping in x" sx := sx + (clipX - destX * xDelta). sy := sy + (clipX - destX * yDelta). ]. noHalftone ifTrue: [halftoneWord := AllOnes] ifFalse: [halftoneWord := self halftoneAt: dy+i-1]. destMask := mask1. nPix := startBits. "Here is the inner loop..." words := nWords. ["pick up word" smoothingCount = 1 ifTrue:["Faster if not smoothing" skewWord := self warpPickSourcePixels: nPix xDeltah: xDelta yDeltah: yDelta xDeltav: deltaP12x yDeltav: deltaP12y dstShiftInc: dstShiftInc flags: mapperFlags. ] ifFalse:["more difficult with smoothing" skewWord := self warpPickSmoothPixels: nPix xDeltah: xDelta yDeltah: yDelta xDeltav: deltaP12x yDeltav: deltaP12y sourceMap: sourceMapOop smoothing: smoothingCount dstShiftInc: dstShiftInc. ]. "align next word access to left most pixel" dstBitShift := dstShiftLeft. destMask = AllOnes ifTrue:["avoid read-modify-write" mergeWord := self mergeFn: (skewWord bitAnd: halftoneWord) with: (self dstLongAt: destIndex). self dstLongAt: destIndex put: (destMask bitAnd: mergeWord). ] ifFalse:[ "General version using dest masking" destWord := self dstLongAt: destIndex. mergeWord := self mergeFn: (skewWord bitAnd: halftoneWord) with: (destWord bitAnd: destMask). destWord := (destMask bitAnd: mergeWord) bitOr: (destWord bitAnd: destMask bitInvert32). self dstLongAt: destIndex put: destWord. ]. self incDestIndex: 4. words = 2 "e.g., is the next word the last word?" ifTrue:["set mask for last word in this row" destMask := mask2. nPix := endBits] ifFalse:["use fullword mask for inner loop" destMask := AllOnes. nPix := destPPW]. (words := words - 1) = 0] whileFalse. "--- end of inner loop ---" pAx := pAx + deltaP12x. pAy := pAy + deltaP12y. pBx := pBx + deltaP43x. pBy := pBy + deltaP43y. self incDestIndex: destDelta]! Item was changed: ----- Method: BitBltSimulation>>warpPickSmoothPixels:xDeltah:yDeltah:xDeltav:yDeltav:sourceMap:smoothing:dstShiftInc: (in category 'pixel mapping') ----- warpPickSmoothPixels: nPixels xDeltah: xDeltah yDeltah: yDeltah xDeltav: xDeltav yDeltav: yDeltav sourceMap: sourceMap smoothing: n dstShiftInc: dstShiftInc "Pick n (sub-) pixels from the source form, mapped by sourceMap, average the RGB values, map by colorMap and return the new word. This version is only called from WarpBlt with smoothingCount > 1" | rgb x y a r g b xx yy xdh ydh xdv ydv dstMask destWord i j k nPix | <inline: false> "nope - too much stuff in here" + <var: #rgb type: #'unsigned int'> dstMask := maskTable at: destDepth. destWord := 0. n = 2 "Try avoiding divides for most common n (divide by 2 is generated as shift)" ifTrue:[xdh := xDeltah // 2. ydh := yDeltah // 2. xdv := xDeltav // 2. ydv := yDeltav // 2] ifFalse:[xdh := xDeltah // n. ydh := yDeltah // n. xdv := xDeltav // n. ydv := yDeltav // n]. i := nPixels. [ x := sx. y := sy. a := r := g := b := 0. "Pick and average n*n subpixels" nPix := 0. "actual number of pixels (not clipped and not transparent)" j := n. [ xx := x. yy := y. k := n. [ "get a single subpixel" rgb := self pickWarpPixelAtX: xx y: yy. (combinationRule=25 "PAINT" and: [rgb = 0]) ifFalse:[ "If not clipped and not transparent, then tally rgb values" nPix := nPix + 1. sourceDepth < 16 ifTrue:[ "Get RGBA values from sourcemap table" rgb := self long32At: sourceMap + (rgb << 2). ] ifFalse:["Already in RGB format" sourceDepth = 16 ifTrue:[rgb := self rgbMap16To32: rgb] ifFalse:[rgb := self rgbMap32To32: rgb]]. b := b + (rgb bitAnd: 255). g := g + (rgb >> 8 bitAnd: 255). r := r + (rgb >> 16 bitAnd: 255). a := a + (rgb >> 24)]. xx := xx + xdh. yy := yy + ydh. (k := k - 1) = 0] whileFalse. x := x + xdv. y := y + ydv. (j := j - 1) = 0] whileFalse. (nPix = 0 or: [combinationRule=25 "PAINT" and: [nPix < (n * n // 2)]]) ifTrue:[ rgb := 0 "All pixels were 0, or most were transparent" ] ifFalse:[ "normalize rgba sums" nPix = 4 "Try to avoid divides for most common n" ifTrue:[r := r >> 2. g := g >> 2. b := b >> 2. a := a >> 2] ifFalse:[ r := r // nPix. g := g // nPix. b := b // nPix. a := a // nPix]. rgb := (a << 24) + (r << 16) + (g << 8) + b. "map the pixel" rgb = 0 ifTrue: [ "only generate zero if pixel is really transparent" (r + g + b + a) > 0 ifTrue: [rgb := 1]]. rgb := self mapPixel: rgb flags: cmFlags. ]. "Mix it in" destWord := destWord bitOr: (rgb bitAnd: dstMask) << dstBitShift. dstBitShift := dstBitShift + dstShiftInc. sx := sx + xDeltah. sy := sy + yDeltah. (i := i - 1) = 0] whileFalse. ^destWord !

1 0

VM Maker: VMMaker.oscog-eem.1753.mcz
by commits＠source.squeak.org 31 Mar '16

31 Mar '16

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.1753.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.1753 Author: eem Time: 1 January 1970, 12:22:37.951321 pm UUID: c2577253-d843-4791-871d-f5e8ba2106b9 Ancestors: VMMaker.oscog-cb.1752 Make VMMaker.oscog-cb.1752's changes compilable. Add a class var to hold the number of store trampolines; use 5 trampolines. Initialize the table for simulation in generateObjectRepresentationTrampolines. Declare ceStoreTrampolines conditionally depending on IMMUTABILITY. Avoid an extra register load in the store trampoline generator. =============== Diff against VMMaker.oscog-cb.1752 =============== Item was changed: ----- Method: CogObjectRepresentation class>>numTrampolines (in category 'accessing') ----- numTrampolines + ^1 "ceStoreCheckTrampoline"! - ^1! Item was changed: CogObjectRepresentation subclass: #CogObjectRepresentationForSpur instanceVariableNames: 'ceScheduleScavengeTrampoline ceSmallActiveContextInMethodTrampoline ceSmallActiveContextInBlockTrampoline ceLargeActiveContextInMethodTrampoline ceLargeActiveContextInBlockTrampoline ceStoreCheckContextReceiverTrampoline ceStoreTrampolines' + classVariableNames: 'NumStoreTrampolines' - classVariableNames: '' poolDictionaries: 'VMBytecodeConstants VMSqueakClassIndices' category: 'VMMaker-JIT'! Item was added: + ----- Method: CogObjectRepresentationForSpur class>>declareCVarsIn: (in category 'translation') ----- + declareCVarsIn: aCodeGen + aCodeGen + var: #ceStoreTrampolines + declareC: ('#if IMMUTABILITY\sqInt ceStoreTrampolines[', NumStoreTrampolines printString, '];\#endif') withCRs! Item was added: + ----- Method: CogObjectRepresentationForSpur class>>initialize (in category 'class initialization') ----- + initialize + "Number of trampolines for the combined IMMUTABILITY/store check instance variable store. + Trampolines from 0 to NumStoreTrampolines - 2 are dedicated to an inst var index, the last one is generic." + NumStoreTrampolines := 5! Item was changed: ----- Method: CogObjectRepresentationForSpur class>>numTrampolines (in category 'accessing') ----- numTrampolines + ^super numTrampolines + + 6 "(small,large)x(method,block) context creation, ceStoreCheckContextReceiverTrampoline and ceScheduleScavengeTrampoline" + + ((initializationOptions at: #IMMUTABILITY ifAbsent: [false]) + ifTrue: [NumStoreTrampolines] + ifFalse: [0])! - (initializationOptions at: #IMMUTABILITY ifAbsent: [false]) - ifTrue: [ ^ super numTrampolines + self new numStoreTrampolines + 7 ] - ifFalse: [ ^ super numTrampolines + 6 ]! Item was changed: ----- Method: CogObjectRepresentationForSpur>>genStoreTrampolineCall: (in category 'compile abstract instructions') ----- genStoreTrampolineCall: instVarIndex <inline: true> + instVarIndex >= (NumStoreTrampolines - 1) - instVarIndex < self numStoreTrampolines - ifFalse: - [ cogit MoveCq: instVarIndex R: TempReg. - cogit CallRT: (ceStoreTrampolines at: self numStoreTrampolines) ] ifTrue: + [ cogit MoveCq: instVarIndex R: TempReg. + cogit CallRT: (ceStoreTrampolines at: NumStoreTrampolines - 1) ] + ifFalse: [ cogit CallRT: (ceStoreTrampolines at: instVarIndex) ]. cogit annotateBytecode: cogit Label! Item was changed: ----- Method: CogObjectRepresentationForSpur>>genStoreTrampolineCalled:instVarIndex: (in category 'initialization') ----- genStoreTrampolineCalled: trampolineName instVarIndex: instVarIndex "Convention: - RcvrResultReg holds the object mutated. If immutability failure: - TempReg holds the instance variable index mutated if instVarIndex > numDedicatedStoreTrampoline - ClassReg holds the value to store Registers are not lived across this trampoline as the immutability failure may need new stack frames." | jumpSC | <option: #IMMUTABILITY> <var: #trampolineName type: #'char *'> <var: #jumpSC type: #'AbstractInstruction *'> <inline: false> cogit zeroOpcodeIndex. + "SendNumArgsReg is mutated but we don't care as register are not live across the trampoline. + There is no reason why registers cannot be saved over the remember: call, but since the + immutability check is a suspension point, registers cannot remain live." + jumpSC := self genJumpMutable: ReceiverResultReg scratchReg: SendNumArgsReg. - jumpSC := self genJumpMutable: ReceiverResultReg scratchReg: SendNumArgsReg. "SendNumArgsReg is mutated but we don't care as register are not live across the trampoline" - instVarIndex = self numStoreTrampolines - ifFalse: [ cogit MoveCq: instVarIndex R: TempReg ]. cogit compileTrampolineFor: #ceCannotAssignTo:withIndex:valueToAssign: numArgs: 3 arg: ReceiverResultReg + arg: (instVarIndex < (NumStoreTrampolines - 1) + ifTrue: [cogit trampolineArgConstant: instVarIndex] + ifFalse: [TempReg]) - arg: TempReg arg: ClassReg arg: nil regsToSave: cogit emptyRegisterMask pushLinkReg: true resultReg: NoReg. "Store check" jumpSC jmpTarget: cogit Label. ^ cogit genTrampolineFor: #remember: called: trampolineName numArgs: 1 arg: ReceiverResultReg arg: nil arg: nil arg: nil regsToSave: cogit emptyRegisterMask pushLinkReg: true resultReg: NoReg appendOpcodes: true! Item was changed: ----- Method: CogObjectRepresentationForSpur>>generateObjectRepresentationTrampolines (in category 'initialization') ----- generateObjectRepresentationTrampolines "Do the store check. Answer the argument for the benefit of the code generator; ReceiverResultReg may be caller-saved and hence smashed by this call. Answering it allows the code generator to reload ReceiverResultReg cheaply. In Spur the only thing we leave to the run-time is adding the receiver to the remembered set and setting its isRemembered bit." self cppIf: IMMUTABILITY ifTrue: + [self cCode: [] inSmalltalk: + [ceStoreTrampolines := CArrayAccessor on: (Array new: NumStoreTrampolines)]. + 0 to: NumStoreTrampolines - 1 do: + [:instVarIndex | + ceStoreTrampolines + at: instVarIndex + put: (self - [ 0 to: self numStoreTrampolines do: - [ :instVarIndex | - ceStoreTrampolines - at: instVarIndex - put: (self genStoreTrampolineCalled: (cogit + trampolineName: 'ceStoreTrampoline' + numArgs: instVarIndex + limit: NumStoreTrampolines - 2) + instVarIndex: instVarIndex)]]. - trampolineName: 'ceStoreTrampoline' - numArgs: instVarIndex - limit: self numStoreTrampolines - 1) - instVarIndex: instVarIndex ) ] ]. ceStoreCheckTrampoline := cogit genTrampolineFor: #remember: called: 'ceStoreCheckTrampoline' arg: ReceiverResultReg regsToSave: (cogit callerSavedRegMask bitClear: (cogit registerMaskFor: ReceiverResultReg)) result: cogit returnRegForStoreCheck. ceStoreCheckContextReceiverTrampoline := self genStoreCheckContextReceiverTrampoline. ceScheduleScavengeTrampoline := cogit genTrampolineFor: #ceScheduleScavenge called: 'ceScheduleScavengeTrampoline' regsToSave: cogit callerSavedRegMask. ceSmallActiveContextInMethodTrampoline := self genActiveContextTrampolineLarge: false inBlock: false called: 'ceSmallMethodContext'. ceSmallActiveContextInBlockTrampoline := self genActiveContextTrampolineLarge: false inBlock: true called: 'ceSmallBlockContext'. ceLargeActiveContextInMethodTrampoline := self genActiveContextTrampolineLarge: true inBlock: false called: 'ceLargeMethodContext'. ceLargeActiveContextInBlockTrampoline := self genActiveContextTrampolineLarge: true inBlock: true called: 'ceLargeBlockContext'! Item was removed: - ----- Method: CogObjectRepresentationForSpur>>numStoreTrampolines (in category 'initialization') ----- - numStoreTrampolines - "Number of trampolines for instance variable store. Trampolines from 0 to - numStoreTrampoline - 1 are dedicated to an inst var index, the last one is generic. - - WARNING: Check the C declaration of ceStoreTrampolines if this number is increased." - ^ 3! Item was removed: - ----- Method: CogObjectRepresentationForSpur>>setCogit:methodZone: (in category 'in-line cacheing') ----- - setCogit: aCogit methodZone: aMethodZone - <doNotGenerate> - super setCogit: aCogit methodZone: aMethodZone. - ceStoreTrampolines := CArrayAccessor on: (Array new: self numStoreTrampolines + 1).! Item was changed: ----- Method: StackToRegisterMappingCogit class>>declareCVarsIn: (in category 'translation') ----- declareCVarsIn: aCodeGen aCodeGen - var: #ceStoreTrampolines - declareC: 'sqint ceStoreTrampolines[4]'; var: #methodAbortTrampolines declareC: 'sqInt methodAbortTrampolines[4]'; var: #picAbortTrampolines declareC: 'sqInt picAbortTrampolines[4]'; var: #picMissTrampolines declareC: 'sqInt picMissTrampolines[4]'; var: 'ceCall0ArgsPIC' declareC: 'void (*ceCall0ArgsPIC)(void)'; var: 'ceCall1ArgsPIC' declareC: 'void (*ceCall1ArgsPIC)(void)'; var: 'ceCall2ArgsPIC' declareC: 'void (*ceCall2ArgsPIC)(void)'; var: #ceCallCogCodePopReceiverArg0Regs declareC: 'void (*ceCallCogCodePopReceiverArg0Regs)(void)'; var: #realCECallCogCodePopReceiverArg0Regs declareC: 'void (*realCECallCogCodePopReceiverArg0Regs)(void)'; var: #ceCallCogCodePopReceiverArg1Arg0Regs declareC: 'void (*ceCallCogCodePopReceiverArg1Arg0Regs)(void)'; var: #realCECallCogCodePopReceiverArg1Arg0Regs declareC: 'void (*realCECallCogCodePopReceiverArg1Arg0Regs)(void)'; var: 'simStack' declareC: 'CogSimStackEntry simStack[', ((CoInterpreter bindingOf: #LargeContextSlots) value * 5 // 4) asString, ']'; var: 'simSelf' type: #CogSimStackEntry; var: #optStatus type: #CogSSOptStatus; var: 'prevBCDescriptor' type: #'BytecodeDescriptor *'. self numPushNilsFunction ifNotNil: [aCodeGen var: 'numPushNilsFunction' declareC: 'sqInt (* const numPushNilsFunction)(struct _BytecodeDescriptor *,sqInt,sqInt,sqInt) = ', (aCodeGen cFunctionNameFor: self numPushNilsFunction); var: 'pushNilSizeFunction' declareC: 'sqInt (* const pushNilSizeFunction)(sqInt,sqInt) = ', (aCodeGen cFunctionNameFor: self pushNilSizeFunction)]. aCodeGen addSelectorTranslation: #register to: (aCodeGen cFunctionNameFor: 'registerr'); addSelectorTranslation: #register: to: (aCodeGen cFunctionNameFor: 'registerr:')!

1 0

VM Maker: VMMaker.oscog-cb.1752.mcz
by commits＠source.squeak.org 31 Mar '16

31 Mar '16

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-cb.1752.mcz ==================== Summary ==================== Name: VMMaker.oscog-cb.1752 Author: cb Time: 31 March 2016, 5:23:26.805 pm UUID: f33aacb8-6e6e-4aae-a6dc-18f006b76afd Ancestors: VMMaker.oscog-eem.1751 Attempt to make multiple trampolines instead of 1 for the store trampolines. By changing in 1 place the code (numStoreTrampolines), choose how many trampolines you want. =============== Diff against VMMaker.oscog-eem.1751 =============== Item was changed: CogObjectRepresentation subclass: #CogObjectRepresentationForSpur + instanceVariableNames: 'ceScheduleScavengeTrampoline ceSmallActiveContextInMethodTrampoline ceSmallActiveContextInBlockTrampoline ceLargeActiveContextInMethodTrampoline ceLargeActiveContextInBlockTrampoline ceStoreCheckContextReceiverTrampoline ceStoreTrampolines' - instanceVariableNames: 'ceScheduleScavengeTrampoline ceSmallActiveContextInMethodTrampoline ceSmallActiveContextInBlockTrampoline ceLargeActiveContextInMethodTrampoline ceLargeActiveContextInBlockTrampoline ceStoreCheckContextReceiverTrampoline ceStoreTrampoline' classVariableNames: '' poolDictionaries: 'VMBytecodeConstants VMSqueakClassIndices' category: 'VMMaker-JIT'! Item was changed: ----- Method: CogObjectRepresentationForSpur class>>numTrampolines (in category 'accessing') ----- numTrampolines (initializationOptions at: #IMMUTABILITY ifAbsent: [false]) + ifTrue: [ ^ super numTrampolines + self new numStoreTrampolines + 7 ] - ifTrue: [ ^ super numTrampolines + 7 ] ifFalse: [ ^ super numTrampolines + 6 ]! Item was added: + ----- Method: CogObjectRepresentationForSpur>>genStoreTrampolineCall: (in category 'compile abstract instructions') ----- + genStoreTrampolineCall: instVarIndex + <inline: true> + instVarIndex < self numStoreTrampolines + ifFalse: + [ cogit MoveCq: instVarIndex R: TempReg. + cogit CallRT: (ceStoreTrampolines at: self numStoreTrampolines) ] + ifTrue: + [ cogit CallRT: (ceStoreTrampolines at: instVarIndex) ]. + cogit annotateBytecode: cogit Label! Item was removed: - ----- Method: CogObjectRepresentationForSpur>>genStoreTrampolineCalled: (in category 'initialization') ----- - genStoreTrampolineCalled: trampolineName - "This can be entered in one of two states, depending on TempReg. - TempReg = 0 => store check - TempReg > 0 => immutability failure - TempReg holds index + 1 in this case as the value 0 is reserved for store checks. - In addition the 0 value is convenient to save one instruction for store checks." - | jumpSC | - <option: #IMMUTABILITY> - <var: #trampolineName type: #'char *'> - <var: #jumpSC type: #'AbstractInstruction *'> - <inline: false> - cogit zeroOpcodeIndex. - cogit PushR: SendNumArgsReg. - jumpSC := self genJumpMutable: ReceiverResultReg scratchReg: SendNumArgsReg. - cogit PopR: SendNumArgsReg. - cogit - compileTrampolineFor: #ceCannotAssignTo:withIndex:valueToAssign: - numArgs: 3 - arg: ReceiverResultReg - arg: TempReg - arg: ClassReg - arg: nil - regsToSave: cogit emptyRegisterMask - pushLinkReg: true - resultReg: NoReg. - - "Store check" - jumpSC jmpTarget: (cogit PopR: SendNumArgsReg). - ^ cogit genTrampolineFor: #remember: - called: trampolineName - numArgs: 1 - arg: ReceiverResultReg - arg: nil - arg: nil - arg: nil - regsToSave: cogit emptyRegisterMask - pushLinkReg: true - resultReg: NoReg - appendOpcodes: true! Item was added: + ----- Method: CogObjectRepresentationForSpur>>genStoreTrampolineCalled:instVarIndex: (in category 'initialization') ----- + genStoreTrampolineCalled: trampolineName instVarIndex: instVarIndex + "Convention: + - RcvrResultReg holds the object mutated. + If immutability failure: + - TempReg holds the instance variable index mutated + if instVarIndex > numDedicatedStoreTrampoline + - ClassReg holds the value to store + Registers are not lived across this trampoline as the + immutability failure may need new stack frames." + + | jumpSC | + <option: #IMMUTABILITY> + <var: #trampolineName type: #'char *'> + <var: #jumpSC type: #'AbstractInstruction *'> + <inline: false> + cogit zeroOpcodeIndex. + jumpSC := self genJumpMutable: ReceiverResultReg scratchReg: SendNumArgsReg. "SendNumArgsReg is mutated but we don't care as register are not live across the trampoline" + instVarIndex = self numStoreTrampolines + ifFalse: [ cogit MoveCq: instVarIndex R: TempReg ]. + cogit + compileTrampolineFor: #ceCannotAssignTo:withIndex:valueToAssign: + numArgs: 3 + arg: ReceiverResultReg + arg: TempReg + arg: ClassReg + arg: nil + regsToSave: cogit emptyRegisterMask + pushLinkReg: true + resultReg: NoReg. + + "Store check" + jumpSC jmpTarget: cogit Label. + ^ cogit genTrampolineFor: #remember: + called: trampolineName + numArgs: 1 + arg: ReceiverResultReg + arg: nil + arg: nil + arg: nil + regsToSave: cogit emptyRegisterMask + pushLinkReg: true + resultReg: NoReg + appendOpcodes: true! Item was changed: ----- Method: CogObjectRepresentationForSpur>>genStoreWithImmutabilityAndStoreCheckSourceReg:slotIndex:destReg:scratchReg:needRestoreRcvr: (in category 'compile abstract instructions') ----- genStoreWithImmutabilityAndStoreCheckSourceReg: sourceReg slotIndex: index destReg: destReg scratchReg: scratchReg needRestoreRcvr: needRestoreRcvr "Store check code is duplicated to use a single trampoline" | immutableJump jmpImmediate jmpDestYoung jmpSourceOld rememberedBitByteOffset jmpAlreadyRemembered mask | <var: #immutableJump type: #'AbstractInstruction *'> <var: #jmpImmediate type: #'AbstractInstruction *'> <var: #jmpDestYoung type: #'AbstractInstruction *'> <var: #jmpSourceOld type: #'AbstractInstruction *'> <var: #jmpAlreadyRemembered type: #'AbstractInstruction *'> immutableJump := self genJumpImmutable: destReg scratchReg: scratchReg. cogit genTraceStores. "do the store" cogit MoveR: sourceReg Mw: index * objectMemory wordSize + objectMemory baseHeaderSize r: destReg. "store check" jmpImmediate := self genJumpImmediate: sourceReg. "Get the old/new boundary in scratchReg" cogit MoveCw: objectMemory storeCheckBoundary R: scratchReg. "Is target young? If so we're done" cogit CmpR: scratchReg R: destReg. "N.B. FLAGS := destReg - scratchReg" jmpDestYoung := cogit JumpBelow: 0. "Is value stored old? If so we're done." cogit CmpR: scratchReg R: sourceReg. "N.B. FLAGS := valueReg - scratchReg" jmpSourceOld := cogit JumpAboveOrEqual: 0. "value is young and target is old. Need to remember this only if the remembered bit is not already set. Test the remembered bit. Only need to fetch the byte containing it, which reduces the size of the mask constant." rememberedBitByteOffset := jmpSourceOld isBigEndian ifTrue: [objectMemory baseHeaderSize - 1 - (objectMemory rememberedBitShift // 8)] ifFalse:[objectMemory rememberedBitShift // 8]. mask := 1 << (objectMemory rememberedBitShift \\ 8). cogit MoveMb: rememberedBitByteOffset r: destReg R: scratchReg. cogit AndCq: mask R: scratchReg. jmpAlreadyRemembered := cogit JumpNonZero: 0. "Set the inst var index for the benefit of the immutability check. The trampoline will repeat the check to choose between the immutbality violation and the store check." + immutableJump jmpTarget: cogit Label. + self genStoreTrampolineCall: index. - immutableJump jmpTarget: (cogit MoveCq: index R: scratchReg). - cogit CallRT: ceStoreTrampoline. - cogit annotateBytecode: cogit Label. needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. jmpImmediate jmpTarget: (jmpDestYoung jmpTarget: (jmpSourceOld jmpTarget: (jmpAlreadyRemembered jmpTarget: cogit Label))). ^ 0! Item was changed: ----- Method: CogObjectRepresentationForSpur>>genStoreWithImmutabilityButNoStoreCheckSourceReg:slotIndex:destReg:scratchReg:needRestoreRcvr: (in category 'compile abstract instructions') ----- genStoreWithImmutabilityButNoStoreCheckSourceReg: sourceReg slotIndex: index destReg: destReg scratchReg: scratchReg needRestoreRcvr: needRestoreRcvr <var: #mutableJump type: #'AbstractInstruction *'> <var: #immutabilityFailure type: #'AbstractInstruction *'> | immutabilityFailure mutableJump | "imm check has its own trampoline" mutableJump := self genJumpMutable: destReg scratchReg: scratchReg. + self genStoreTrampolineCall: index. - cogit MoveCq: index R: TempReg. - cogit CallRT: ceStoreTrampoline. - cogit annotateBytecode: cogit Label. needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. immutabilityFailure := cogit Jump: 0. mutableJump jmpTarget: cogit Label. cogit genTraceStores. "do the store" cogit MoveR: sourceReg Mw: index * objectMemory wordSize + objectMemory baseHeaderSize r: destReg. immutabilityFailure jmpTarget: cogit Label. ^ 0! Item was changed: ----- Method: CogObjectRepresentationForSpur>>generateObjectRepresentationTrampolines (in category 'initialization') ----- generateObjectRepresentationTrampolines "Do the store check. Answer the argument for the benefit of the code generator; ReceiverResultReg may be caller-saved and hence smashed by this call. Answering it allows the code generator to reload ReceiverResultReg cheaply. In Spur the only thing we leave to the run-time is adding the receiver to the remembered set and setting its isRemembered bit." self cppIf: IMMUTABILITY + ifTrue: + [ 0 to: self numStoreTrampolines do: + [ :instVarIndex | + ceStoreTrampolines + at: instVarIndex + put: (self + genStoreTrampolineCalled: (cogit + trampolineName: 'ceStoreTrampoline' + numArgs: instVarIndex + limit: self numStoreTrampolines - 1) + instVarIndex: instVarIndex ) ] ]. - ifTrue: [ceStoreTrampoline := self genStoreTrampolineCalled: 'ceStoreTrampoline']. ceStoreCheckTrampoline := cogit genTrampolineFor: #remember: called: 'ceStoreCheckTrampoline' arg: ReceiverResultReg regsToSave: (cogit callerSavedRegMask bitClear: (cogit registerMaskFor: ReceiverResultReg)) result: cogit returnRegForStoreCheck. ceStoreCheckContextReceiverTrampoline := self genStoreCheckContextReceiverTrampoline. ceScheduleScavengeTrampoline := cogit genTrampolineFor: #ceScheduleScavenge called: 'ceScheduleScavengeTrampoline' regsToSave: cogit callerSavedRegMask. ceSmallActiveContextInMethodTrampoline := self genActiveContextTrampolineLarge: false inBlock: false called: 'ceSmallMethodContext'. ceSmallActiveContextInBlockTrampoline := self genActiveContextTrampolineLarge: false inBlock: true called: 'ceSmallBlockContext'. ceLargeActiveContextInMethodTrampoline := self genActiveContextTrampolineLarge: true inBlock: false called: 'ceLargeMethodContext'. ceLargeActiveContextInBlockTrampoline := self genActiveContextTrampolineLarge: true inBlock: true called: 'ceLargeBlockContext'! Item was added: + ----- Method: CogObjectRepresentationForSpur>>numStoreTrampolines (in category 'initialization') ----- + numStoreTrampolines + "Number of trampolines for instance variable store. Trampolines from 0 to + numStoreTrampoline - 1 are dedicated to an inst var index, the last one is generic. + + WARNING: Check the C declaration of ceStoreTrampolines if this number is increased." + ^ 3! Item was added: + ----- Method: CogObjectRepresentationForSpur>>setCogit:methodZone: (in category 'in-line cacheing') ----- + setCogit: aCogit methodZone: aMethodZone + <doNotGenerate> + super setCogit: aCogit methodZone: aMethodZone. + ceStoreTrampolines := CArrayAccessor on: (Array new: self numStoreTrampolines + 1).! Item was changed: ----- Method: StackToRegisterMappingCogit class>>declareCVarsIn: (in category 'translation') ----- declareCVarsIn: aCodeGen aCodeGen + var: #ceStoreTrampolines + declareC: 'sqint ceStoreTrampolines[4]'; var: #methodAbortTrampolines declareC: 'sqInt methodAbortTrampolines[4]'; var: #picAbortTrampolines declareC: 'sqInt picAbortTrampolines[4]'; var: #picMissTrampolines declareC: 'sqInt picMissTrampolines[4]'; var: 'ceCall0ArgsPIC' declareC: 'void (*ceCall0ArgsPIC)(void)'; var: 'ceCall1ArgsPIC' declareC: 'void (*ceCall1ArgsPIC)(void)'; var: 'ceCall2ArgsPIC' declareC: 'void (*ceCall2ArgsPIC)(void)'; var: #ceCallCogCodePopReceiverArg0Regs declareC: 'void (*ceCallCogCodePopReceiverArg0Regs)(void)'; var: #realCECallCogCodePopReceiverArg0Regs declareC: 'void (*realCECallCogCodePopReceiverArg0Regs)(void)'; var: #ceCallCogCodePopReceiverArg1Arg0Regs declareC: 'void (*ceCallCogCodePopReceiverArg1Arg0Regs)(void)'; var: #realCECallCogCodePopReceiverArg1Arg0Regs declareC: 'void (*realCECallCogCodePopReceiverArg1Arg0Regs)(void)'; var: 'simStack' declareC: 'CogSimStackEntry simStack[', ((CoInterpreter bindingOf: #LargeContextSlots) value * 5 // 4) asString, ']'; var: 'simSelf' type: #CogSimStackEntry; var: #optStatus type: #CogSSOptStatus; var: 'prevBCDescriptor' type: #'BytecodeDescriptor *'. self numPushNilsFunction ifNotNil: [aCodeGen var: 'numPushNilsFunction' declareC: 'sqInt (* const numPushNilsFunction)(struct _BytecodeDescriptor *,sqInt,sqInt,sqInt) = ', (aCodeGen cFunctionNameFor: self numPushNilsFunction); var: 'pushNilSizeFunction' declareC: 'sqInt (* const pushNilSizeFunction)(sqInt,sqInt) = ', (aCodeGen cFunctionNameFor: self pushNilSizeFunction)]. aCodeGen addSelectorTranslation: #register to: (aCodeGen cFunctionNameFor: 'registerr'); addSelectorTranslation: #register: to: (aCodeGen cFunctionNameFor: 'registerr:')!

1 0

Re: [Vm-dev] VM Maker: VMMaker.oscog-cb.1748.mcz
by Clément Bera 31 Mar '16

31 Mar '16

It saves 2 bytes for inst var 0 and 5 bytes for inst var between 1 and 255 in x86. There is one less branch to take on the common path too. Not sure if this matters a lot. The difference should be better on x64 as in x86 MoveCq:R: calls MoveCw:R: if the quick constant is not 0, and we have many little constants more than 0. Isn't there a better way to move quick constant to register on x86 than the full word constant ? Maybe we should change MoveCq:R: on x86 to do a xor reg,reg, movq r8 as suggested in the comment. 2016-03-30 15:52 GMT+02:00 <commits(a)source.squeak.org>: > > Eliot Miranda uploaded a new version of VMMaker to project VM Maker: > http://source.squeak.org/VMMaker/VMMaker.oscog-cb.1748.mcz > > ==================== Summary ==================== > > Name: VMMaker.oscog-cb.1748 > Author: cb > Time: 30 March 2016, 3:52:01.732 pm > UUID: db420bcd-e4aa-4cad-9543-047274e49915 > Ancestors: VMMaker.oscog-nice.1747 > > Reworked machine code generation of immutability so for common stores it > uses a single trampoline for both store checks and immutability checks. > > I have simulation bug due to large integers, so I am not entirely sure > everything is working, but generated code looks good. > > =============== Diff against VMMaker.oscog-nice.1747 =============== > > Item was changed: > ----- Method: > CogObjectRepresentation>>genStoreSourceReg:slotIndex:destReg:scratchReg:inFrame: > (in category 'compile abstract instructions') ----- > genStoreSourceReg: sourceReg slotIndex: index destReg: destReg > scratchReg: scratchReg inFrame: inFrame > + <inline: true> > + self > + cppIf: IMMUTABILITY > + ifTrue: > + [ ^ self > + genStoreWithImmutabilityCheckSourceReg: > sourceReg > + slotIndex: index > + destReg: destReg > + scratchReg: scratchReg > + needsStoreCheck: true > + needRestoreRcvr: false "RcvrResultReg > doesn't need to be live across the instructions" ] > + ifFalse: > + [ ^ self > + genStoreSourceReg: sourceReg > + slotIndex: index > + destReg: destReg > + scratchReg: scratchReg > + inFrame: inFrame > + needsStoreCheck: true ]! > - ^ self > - genStoreSourceReg: sourceReg > - slotIndex: index > - destReg: destReg > - scratchReg: scratchReg > - inFrame: inFrame > - needsStoreCheck: true! > > Item was changed: > CogObjectRepresentation subclass: #CogObjectRepresentationForSpur > + instanceVariableNames: 'ceScheduleScavengeTrampoline > ceSmallActiveContextInMethodTrampoline > ceSmallActiveContextInBlockTrampoline > ceLargeActiveContextInMethodTrampoline > ceLargeActiveContextInBlockTrampoline ceStoreCheckContextReceiverTrampoline > ceStoreTrampoline' > - instanceVariableNames: 'ceScheduleScavengeTrampoline > ceSmallActiveContextInMethodTrampoline > ceSmallActiveContextInBlockTrampoline > ceLargeActiveContextInMethodTrampoline > ceLargeActiveContextInBlockTrampoline ceStoreCheckContextReceiverTrampoline > ceCannotAssignToWithIndexTrampoline' > classVariableNames: '' > poolDictionaries: 'VMBytecodeConstants VMSqueakClassIndices' > category: 'VMMaker-JIT'! > > Item was removed: > - ----- Method: > CogObjectRepresentationForSpur>>genImmutableCheck:slotIndex:sourceReg:scratchReg:needRestoreRcvr: > (in category 'compile abstract instructions') ----- > - genImmutableCheck: regHoldingObjectMutated slotIndex: index sourceReg: > regHoldingValueToStore scratchReg: scratchReg needRestoreRcvr: > needRestoreRcvr > - | mutableJump fail | > - <var: #mutableJump type: #'AbstractInstruction *'> > - <var: #fail type: #'AbstractInstruction *'> > - <inline: true> > - <option: #IMMUTABILITY> > - "Trampoline convention: > - - objectMutated passed in ReceiverResultReg > - - index (unboxed) passed in TempReg > - - valueToStore passed in ClassReg. > - Simulated stack is flushed, but if needRestoreRcvr is true > - the receiver has to be live after this operation." > - self assert: regHoldingObjectMutated == ReceiverResultReg. > - self assert: scratchReg == TempReg. > - self assert: regHoldingValueToStore == ClassReg. > - mutableJump := self genJumpMutable: ReceiverResultReg scratchReg: > TempReg. > - > - "We reach this code if the object mutated is immutable." > - cogit MoveCq: index R: TempReg. > - "trampoline call and mcpc to bcpc annotation." > - cogit CallRT: ceCannotAssignToWithIndexTrampoline. > - cogit annotateBytecode: cogit Label. > - "restore ReceiverResultReg state if needed, the rest of the state > is spilled" > - needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. > - fail := cogit Jump: 0. > - > - "We reach this code is the object mutated is mutable" > - mutableJump jmpTarget: cogit Label. > - > - ^ fail! > > Item was changed: > ----- Method: > CogObjectRepresentationForSpur>>genJumpBaseHeaderImmutable: (in category > 'compile abstract instructions') ----- > genJumpBaseHeaderImmutable: baseHeaderReg > "baseHeader holds at least the least significant 32 bits of the > object" > <returnTypeC: #'AbstractInstruction *'> > <option: #IMMUTABILITY> > + <inline: true> > cogit TstCq: objectMemory immutableBitMask R: baseHeaderReg. > ^ cogit JumpNonZero: 0! > > Item was changed: > ----- Method: CogObjectRepresentationForSpur>>genJumpBaseHeaderMutable: > (in category 'compile abstract instructions') ----- > genJumpBaseHeaderMutable: baseHeaderReg > "baseHeader holds at least the least significant 32 bits of the > object" > <returnTypeC: #'AbstractInstruction *'> > <option: #IMMUTABILITY> > + <inline: true> > cogit TstCq: objectMemory immutableBitMask R: baseHeaderReg. > ^ cogit JumpZero: 0! > > Item was added: > + ----- Method: > CogObjectRepresentationForSpur>>genJumpImmutable:scratchReg: (in category > 'compile abstract instructions') ----- > + genJumpImmutable: sourceReg scratchReg: scratchReg > + <returnTypeC: #'AbstractInstruction *'> > + <option: #IMMUTABILITY> > + cogit MoveMw: 0 r: sourceReg R: scratchReg. > + ^ self genJumpBaseHeaderImmutable: scratchReg! > > Item was changed: > ----- Method: > CogObjectRepresentationForSpur>>genStoreSourceReg:slotIndex:destReg:scratchReg:inFrame:needsStoreCheck: > (in category 'compile abstract instructions') ----- > genStoreSourceReg: sourceReg slotIndex: index destReg: destReg > scratchReg: scratchReg inFrame: inFrame needsStoreCheck: needsStoreCheck > + > + cogit genTraceStores. > "do the store" > cogit MoveR: sourceReg > Mw: index * objectMemory wordSize + objectMemory > baseHeaderSize > r: destReg. > "now the check. needStoreCheck is false if the JIT has figured out > that the value stored does not need the check (immediate, nil, true, false)" > needsStoreCheck ifTrue: > [ ^ self > genStoreCheckReceiverReg: destReg > valueReg: sourceReg > scratchReg: scratchReg > inFrame: inFrame ]. > ^ 0! > > Item was added: > + ----- Method: CogObjectRepresentationForSpur>>genStoreTrampolineCalled: > (in category 'initialization') ----- > + genStoreTrampolineCalled: trampolineName > + "This can be entered in one of two states, depending on TempReg. > + TempReg = 0 => store check > + TempReg > 0 => immutability failure > + TempReg holds index + 1 in this case as the value 0 is reserved > for store checks. > + In addition the 0 value is convenient to save one instruction for > store checks." > + | jumpSC | > + <var: #trampolineName type: #'char *'> > + <var: #jumpSC type: #'AbstractInstruction *'> > + <inline: false> > + cogit zeroOpcodeIndex. > + cogit CmpCq: 0 R: TempReg. > + jumpSC := cogit JumpZero: 0. > + > + "CannotAssignTo:, we restore the index." > + cogit SubCq: 1 R: TempReg. > + cogit > + compileTrampolineFor: > #ceCannotAssignTo:withIndex:valueToAssign: > + numArgs: 3 > + arg: ReceiverResultReg > + arg: TempReg > + arg: ClassReg > + arg: nil > + regsToSave: cogit emptyRegisterMask > + pushLinkReg: true > + resultReg: NoReg. > + > + "Store check" > + jumpSC jmpTarget: cogit Label. > + ^ cogit genTrampolineFor: #remember: > + called: trampolineName > + numArgs: 1 > + arg: ReceiverResultReg > + arg: nil > + arg: nil > + arg: nil > + regsToSave: cogit emptyRegisterMask > + pushLinkReg: true > + resultReg: NoReg > + appendOpcodes: true! > > Item was added: > + ----- Method: > CogObjectRepresentationForSpur>>genStoreWithImmutabilityAndStoreCheckSourceReg:slotIndex:destReg:scratchReg:needRestoreRcvr: > (in category 'compile abstract instructions') ----- > + genStoreWithImmutabilityAndStoreCheckSourceReg: sourceReg slotIndex: > index destReg: destReg scratchReg: scratchReg needRestoreRcvr: > needRestoreRcvr > + "Store check code is duplicated to use a single trampoline" > + <var: #immutableJump type: #'AbstractInstruction *'> > + <var: #trampJump type: #'AbstractInstruction *'> > + <var: #jmpImmediate type: #'AbstractInstruction *'> > + <var: #jmpDestYoung type: #'AbstractInstruction *'> > + <var: #jmpSourceOld type: #'AbstractInstruction *'> > + <var: #jmpAlreadyRemembered type: #'AbstractInstruction *'> > + | immutableJump trampJump jmpImmediate jmpDestYoung jmpSourceOld > rememberedBitByteOffset jmpAlreadyRemembered mask | > + > + immutableJump := self genJumpImmutable: destReg scratchReg: > scratchReg. > + > + cogit genTraceStores. > + > + "do the store" > + cogit MoveR: sourceReg > + Mw: index * objectMemory wordSize + objectMemory > baseHeaderSize > + r: destReg. > + > + "store check" > + jmpImmediate := self genJumpImmediate: sourceReg. > + "Get the old/new boundary in scratchReg" > + cogit MoveCw: objectMemory storeCheckBoundary R: scratchReg. > + "Is target young? If so we're done" > + cogit CmpR: scratchReg R: destReg. "N.B. FLAGS := destReg - > scratchReg" > + jmpDestYoung := cogit JumpBelow: 0. > + "Is value stored old? If so we're done." > + cogit CmpR: scratchReg R: sourceReg. "N.B. FLAGS := valueReg - > scratchReg" > + jmpSourceOld := cogit JumpAboveOrEqual: 0. > + "value is young and target is old. > + Need to remember this only if the remembered bit is not already > set. > + Test the remembered bit. Only need to fetch the byte containing > it, > + which reduces the size of the mask constant." > + rememberedBitByteOffset := jmpSourceOld isBigEndian > + > ifTrue: [objectMemory baseHeaderSize - 1 - (objectMemory > rememberedBitShift // 8)] > + > ifFalse:[objectMemory rememberedBitShift // 8]. > + mask := 1 << (objectMemory rememberedBitShift \\ 8). > + cogit MoveMb: rememberedBitByteOffset r: destReg R: scratchReg. > + cogit AndCq: mask R: scratchReg. > + jmpAlreadyRemembered := cogit JumpNonZero: 0. > + "We know scratchReg now holds 0, this is convenient because the > trampoline > + convention expects 0 for store check in scratchReg. What a > coincidence ;-)" > + "Remembered bit is not set. Call store check to insert dest into > remembered table." > + trampJump := cogit Jump: 0. > + "Here we reach the trampoline for Immutability failure" > + immutableJump jmpTarget: (cogit MoveCq: index + 1 R: scratchReg). > "index + 1 as 0 is reserved for store checks" > + trampJump jmpTarget: (cogit CallRT: ceStoreTrampoline). > + cogit annotateBytecode: cogit Label. > + needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. > + > + jmpImmediate jmpTarget: > + (jmpDestYoung jmpTarget: > + (jmpSourceOld jmpTarget: > + (jmpAlreadyRemembered jmpTarget: > + cogit Label))). > + > + ^ 0! > > Item was added: > + ----- Method: > CogObjectRepresentationForSpur>>genStoreWithImmutabilityButNoStoreCheckSourceReg:slotIndex:destReg:scratchReg:needRestoreRcvr: > (in category 'compile abstract instructions') ----- > + genStoreWithImmutabilityButNoStoreCheckSourceReg: sourceReg slotIndex: > index destReg: destReg scratchReg: scratchReg needRestoreRcvr: > needRestoreRcvr > + > + <var: #immutableJump type: #'AbstractInstruction *'> > + <var: #immutabilityFailure type: #'AbstractInstruction *'> > + | immutabilityFailure mutableJump | > + > + "imm check has its own trampoline" > + mutableJump := self genJumpMutable: destReg scratchReg: scratchReg. > + cogit MoveCq: index + 1 R: TempReg. "index + 1 as 0 is reserved > for store checks" > + cogit CallRT: ceStoreTrampoline. > + cogit annotateBytecode: cogit Label. > + needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. > + immutabilityFailure := cogit Jump: 0. > + mutableJump jmpTarget: cogit Label. > + > + cogit genTraceStores. > + > + "do the store" > + cogit MoveR: sourceReg > + Mw: index * objectMemory wordSize + objectMemory > baseHeaderSize > + r: destReg. > + > + immutabilityFailure jmpTarget: cogit Label. > + > + ^ 0! > > Item was added: > + ----- Method: > CogObjectRepresentationForSpur>>genStoreWithImmutabilityCheckSourceReg:slotIndex:destReg:scratchReg:needsStoreCheck:needRestoreRcvr: > (in category 'compile abstract instructions') ----- > + genStoreWithImmutabilityCheckSourceReg: sourceReg slotIndex: index > destReg: destReg scratchReg: scratchReg needsStoreCheck: needsStoreCheck > needRestoreRcvr: needRestoreRcvr > + "We know there is a frame as immutability check requires a frame" > + "needRestoreRcvr has to be true to keep RcvrResultReg live with > the receiver in it across the trampoline" > + > + "Trampoline convention..." > + self assert: destReg == ReceiverResultReg. > + self assert: scratchReg == TempReg. > + self assert: sourceReg == ClassReg. > + > + needsStoreCheck > + ifTrue: > + [ self > + > genStoreWithImmutabilityAndStoreCheckSourceReg: sourceReg > + slotIndex: index > + destReg: destReg > + scratchReg: scratchReg > + needRestoreRcvr: needRestoreRcvr ] > + ifFalse: > + [ self > + > genStoreWithImmutabilityButNoStoreCheckSourceReg: sourceReg > + slotIndex: index > + destReg: destReg > + scratchReg: scratchReg > + needRestoreRcvr: needRestoreRcvr ]. > + ^ 0! > > Item was changed: > ----- Method: > CogObjectRepresentationForSpur>>generateObjectRepresentationTrampolines (in > category 'initialization') ----- > generateObjectRepresentationTrampolines > "Do the store check. Answer the argument for the benefit of the > code generator; > ReceiverResultReg may be caller-saved and hence smashed by this > call. Answering > it allows the code generator to reload ReceiverResultReg cheaply. > In Spur the only thing we leave to the run-time is adding the > receiver to the > remembered set and setting its isRemembered bit." > self > cppIf: IMMUTABILITY > + ifTrue: [ceStoreTrampoline := self > genStoreTrampolineCalled: 'ceStoreTrampoline']. > - ifTrue: "c.f. > genImmutableCheck:slotIndex:sourceReg:scratchReg:popBoolean:needRestoreRcvr:" > - [ceCannotAssignToWithIndexTrampoline := cogit > - > genTrampolineFor: > #ceCannotAssignTo:withIndex:valueToAssign: > - > called: > 'ceCannotAssignToWithIndexTrampoline' > - > arg: ReceiverResultReg > - > arg: TempReg > - > arg: ClassReg]. > ceStoreCheckTrampoline := cogit > > genTrampolineFor: #remember: > > called: 'ceStoreCheckTrampoline' > > arg: ReceiverResultReg > > regsToSave: (cogit callerSavedRegMask bitClear: (cogit registerMaskFor: > ReceiverResultReg)) > > result: cogit returnRegForStoreCheck. > ceStoreCheckContextReceiverTrampoline := self > genStoreCheckContextReceiverTrampoline. > ceScheduleScavengeTrampoline := cogit > > genTrampolineFor: #ceScheduleScavenge > > called: 'ceScheduleScavengeTrampoline' > > regsToSave: cogit callerSavedRegMask. > ceSmallActiveContextInMethodTrampoline := self > genActiveContextTrampolineLarge: false inBlock: false called: > 'ceSmallMethodContext'. > ceSmallActiveContextInBlockTrampoline := self > genActiveContextTrampolineLarge: false inBlock: true called: > 'ceSmallBlockContext'. > ceLargeActiveContextInMethodTrampoline := self > genActiveContextTrampolineLarge: true inBlock: false called: > 'ceLargeMethodContext'. > ceLargeActiveContextInBlockTrampoline := self > genActiveContextTrampolineLarge: true inBlock: true called: > 'ceLargeBlockContext'! > > Item was changed: > ----- Method: > CogObjectRepresentationForSqueakV3>>genStoreSourceReg:slotIndex:destReg:scratchReg:inFrame:needsStoreCheck: > (in category 'compile abstract instructions') ----- > genStoreSourceReg: sourceReg slotIndex: index destReg: destReg > scratchReg: scratchReg inFrame: inFrame needsStoreCheck: needsStoreCheck > | jmpImmediate jmpDestYoung jmpSourceOld jmpAlreadyRoot mask > rootBitByteOffset | > <var: #jmpImmediate type: #'AbstractInstruction *'> > <var: #jmpDestYoung type: #'AbstractInstruction *'> > <var: #jmpSourceOld type: #'AbstractInstruction *'> > <var: #jmpAlreadyRoot type: #'AbstractInstruction *'> > + > + cogit genTraceStores. > "do the store" > cogit MoveR: sourceReg Mw: index * objectMemory wordSize + > objectMemory baseHeaderSize r: destReg. > "if no need for the store check then returns" > needsStoreCheck ifFalse: [ ^ 0 ]. > "now the check. Is value stored an integer? If so we're done" > jmpImmediate := self genJumpImmediate: sourceReg. > "Get the old/new boundary in scratchReg" > cogit MoveAw: objectMemory youngStartAddress R: scratchReg. > "Is target young? If so we're done" > cogit CmpR: scratchReg R: destReg. "N.B. FLAGS := destReg - > scratchReg" > jmpDestYoung := cogit JumpAboveOrEqual: 0. > "Is value stored old? If so we're done." > cogit CmpR: scratchReg R: sourceReg. "N.B. FLAGS := sourceReg - > scratchReg" > jmpSourceOld := cogit JumpBelow: 0. > "value is young and target is old. > Need to make this a root if the root bit is not already set. > Test the root bit. Only need to fetch the byte containing it, > which reduces the size of the mask constant." > rootBitByteOffset := jmpSourceOld isBigEndian > ifTrue: > [objectMemory wordSize - RootBitDigitLength] > > ifFalse:[RootBitDigitLength - 1]. > mask := RootBitDigitLength > 1 > ifTrue: [RootBit >> (RootBitDigitLength - > 1 * 8)] > ifFalse: [RootBit]. > cogit MoveMb: rootBitByteOffset r: destReg R: scratchReg. > cogit AndCq: mask R: scratchReg. > jmpAlreadyRoot := cogit JumpNonZero: 0. > "Root bit is not set. Call store check to insert dest into root > table." > self assert: destReg == ReceiverResultReg. > cogit > evaluateTrampolineCallBlock: [cogit CallRT: > ceStoreCheckTrampoline] > protectLinkRegIfNot: inFrame. > jmpImmediate jmpTarget: > (jmpDestYoung jmpTarget: > (jmpSourceOld jmpTarget: > (jmpAlreadyRoot jmpTarget: > cogit Label))). > ^0! > > Item was changed: > ----- Method: SimpleStackBasedCogit>>genStorePop:LiteralVariable: (in > category 'bytecode generator support') ----- > genStorePop: popBoolean LiteralVariable: litVarIndex > <inline: false> > + | association | > - | association immutabilityFailure | > - <var: #immutabilityFailure type: #'AbstractInstruction *'> > "The only reason we assert needsFrame here is that in a frameless > method > ReceiverResultReg must and does contain only self, but the > ceStoreCheck > trampoline expects the target of the store to be in > ReceiverResultReg. So > in a frameless method we would have a conflict between the > receiver and > the literal store, unless we we smart enough to realise that > ReceiverResultReg > was unused after the literal variable store, unlikely given that > methods > return self by default." > self assert: needsFrame. > association := self getLiteral: litVarIndex. > self genMoveConstant: association R: ReceiverResultReg. > objectRepresentation > genEnsureObjInRegNotForwarded: ReceiverResultReg > scratchReg: TempReg. > popBoolean > ifTrue: [self PopR: ClassReg] > ifFalse: [self MoveMw: 0 r: SPReg R: ClassReg]. > + self > + genStoreSourceReg: ClassReg > + slotIndex: ValueIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > - traceStores > 0 ifTrue: > - [self CallRT: ceTraceStoreTrampoline]. > - self cppIf: IMMUTABILITY ifTrue: > - [immutabilityFailure := objectRepresentation > - > genImmutableCheck: ReceiverResultReg > - > slotIndex: ValueIndex > - > sourceReg: ClassReg > - > scratchReg: TempReg > - > needRestoreRcvr: true]. > - objectRepresentation > - genStoreSourceReg: ClassReg > - slotIndex: ValueIndex > - destReg: ReceiverResultReg > - scratchReg: TempReg > inFrame: needsFrame. > - > - self cppIf: IMMUTABILITY ifTrue: > - [immutabilityFailure jmpTarget: self Label]. > - > ^0! > > Item was changed: > ----- Method: > SimpleStackBasedCogit>>genStorePop:MaybeContextReceiverVariable: (in > category 'bytecode generator support') ----- > genStorePop: popBoolean MaybeContextReceiverVariable: slotIndex > <inline: false> > + | jmpSingle jmpDone | > - | jmpSingle jmpDone immutabilityFailure | > - <var: #immutabilityFailure type: #'AbstractInstruction *'> > <var: #jmpSingle type: #'AbstractInstruction *'> > <var: #jmpDone type: #'AbstractInstruction *'> > "The reason we need a frame here is that assigning to an inst var > of a context may > involve wholesale reorganization of stack pages, and the only way > to preserve the > execution state of an activation in that case is if it has a > frame." > self assert: needsFrame. > self putSelfInReceiverResultReg. > objectRepresentation > genLoadSlot: SenderIndex > sourceReg: ReceiverResultReg > destReg: TempReg. > self MoveMw: 0 r: SPReg R: ClassReg. > jmpSingle := objectRepresentation > genJumpNotSmallIntegerInScratchReg: TempReg. > self MoveCq: slotIndex R: SendNumArgsReg. > self CallRT: ceStoreContextInstVarTrampoline. > jmpDone := self Jump: 0. > jmpSingle jmpTarget: self Label. > - traceStores > 0 ifTrue: > - [self CallRT: ceTraceStoreTrampoline]. > - self cppIf: IMMUTABILITY ifTrue: > - [immutabilityFailure := objectRepresentation > - > genImmutableCheck: ReceiverResultReg > - > slotIndex: slotIndex > - > sourceReg: ClassReg > - > scratchReg: TempReg > - > needRestoreRcvr: true]. > - objectRepresentation > - genStoreSourceReg: ClassReg > - slotIndex: slotIndex > - destReg: ReceiverResultReg > - scratchReg: TempReg > - inFrame: true. > - jmpDone jmpTarget: self Label. > popBoolean ifTrue: > [self AddCq: objectMemory wordSize R: SPReg]. > + self > + genStoreSourceReg: ClassReg > + slotIndex: slotIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > + inFrame: needsFrame. > + jmpDone jmpTarget: self Label. > - self cppIf: IMMUTABILITY ifTrue: > - [immutabilityFailure jmpTarget: self Label]. > ^0! > > Item was changed: > ----- Method: SimpleStackBasedCogit>>genStorePop:ReceiverVariable: (in > category 'bytecode generator support') ----- > genStorePop: popBoolean ReceiverVariable: slotIndex > <inline: false> > - | immutabilityFailure | > - <var: #immutabilityFailure type: #'AbstractInstruction *'> > needsFrame ifTrue: > [self putSelfInReceiverResultReg]. > popBoolean > ifTrue: [self PopR: ClassReg] > ifFalse: [self MoveMw: 0 r: SPReg R: ClassReg]. > + self > + genStoreSourceReg: ClassReg > + slotIndex: slotIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > - traceStores > 0 ifTrue: > - [self CallRT: ceTraceStoreTrampoline]. > - self cppIf: IMMUTABILITY ifTrue: > - [immutabilityFailure := objectRepresentation > - > genImmutableCheck: ReceiverResultReg > - > slotIndex: slotIndex > - > sourceReg: ClassReg > - > scratchReg: TempReg > - > needRestoreRcvr: true]. > - objectRepresentation > - genStoreSourceReg: ClassReg > - slotIndex: slotIndex > - destReg: ReceiverResultReg > - scratchReg: TempReg > inFrame: needsFrame. > - > - self cppIf: IMMUTABILITY ifTrue: > - [immutabilityFailure jmpTarget: self Label]. > - > ^0! > > Item was changed: > ----- Method: SimpleStackBasedCogit>>genStorePop:RemoteTemp:At: (in > category 'bytecode generator support') ----- > genStorePop: popBoolean RemoteTemp: slotIndex At: remoteTempIndex > <inline: false> > "The only reason we assert needsFrame here is that in a frameless > method > ReceiverResultReg must and does contain only self, but the > ceStoreCheck > trampoline expects the target of the store to be in > ReceiverResultReg. So > in a frameless method we would have a conflict between the > receiver and > the temote temp store, unless we we smart enough to realise that > ReceiverResultReg was unused after the literal variable store, > unlikely given > that methods return self by default." > self assert: needsFrame. > popBoolean > ifTrue: [self PopR: ClassReg] > ifFalse: [self MoveMw: 0 r: SPReg R: ClassReg]. > self MoveMw: (self frameOffsetOfTemporary: remoteTempIndex) r: > FPReg R: ReceiverResultReg. > - traceStores > 0 ifTrue: > - [self CallRT: ceTraceStoreTrampoline]. > ^objectRepresentation > genStoreSourceReg: ClassReg > slotIndex: slotIndex > destReg: ReceiverResultReg > scratchReg: TempReg > inFrame: needsFrame! > > Item was added: > + ----- Method: SimpleStackBasedCogit>>genTraceStores (in category > 'bytecode generator support') ----- > + genTraceStores > + <inline: true> > + traceStores > 0 ifTrue: [ self CallRT: ceTraceStoreTrampoline ].! > > Item was added: > + ----- Method: > StackToRegisterMappingCogit>>genImmutabilityCheckStorePop:LiteralVariable: > (in category 'bytecode generator support') ----- > + genImmutabilityCheckStorePop: popBoolean LiteralVariable: litVarIndex > + <inline: true> > + | association needStoreCheck | > + "The only reason we assert needsFrame here is that in a frameless > method > + ReceiverResultReg must and does contain only self, but the > ceStoreCheck > + trampoline expects the target of the store to be in > ReceiverResultReg. So > + in a frameless method we would have a conflict between the > receiver and > + the literal store, unless we we smart enough to realise that > ReceiverResultReg > + was unused after the literal variable store, unlikely given that > methods > + return self by default." > + self assert: needsFrame. > + "N.B. No need to check the stack for references because we > generate code for > + literal variable loads that stores the result in a register, > deferring only the register push." > + needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > + association := self getLiteral: litVarIndex. > + optStatus isReceiverResultRegLive: false. > + self ssAllocateRequiredReg: ReceiverResultReg. "for store > trampoline call in genStoreSourceReg: has to be ReceiverResultReg" > + self genMoveConstant: association R: ReceiverResultReg. > + objectRepresentation genEnsureObjInRegNotForwarded: > ReceiverResultReg scratchReg: TempReg. > + self ssAllocateRequiredReg: ClassReg. > + self ssStoreAndReplacePop: popBoolean toReg: ClassReg. > + self ssFlushTo: simStackPtr. > + objectRepresentation > + genStoreWithImmutabilityCheckSourceReg: ClassReg > + slotIndex: ValueIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > + needsStoreCheck: needStoreCheck > + needRestoreRcvr: false. > + ^ 0! > > Item was added: > + ----- Method: > StackToRegisterMappingCogit>>genImmutabilityCheckStorePop:MaybeContextReceiverVariable: > (in category 'bytecode generator support') ----- > + genImmutabilityCheckStorePop: popBoolean MaybeContextReceiverVariable: > slotIndex > + <inline: true> > + | jmpSingle jmpDone needStoreCheck | > + <var: #jmpSingle type: #'AbstractInstruction *'> > + <var: #jmpDone type: #'AbstractInstruction *'> > + "The reason we need a frame here is that assigning to an inst var > of a context may > + involve wholesale reorganization of stack pages, and the only way > to preserve the > + execution state of an activation in that case is if it has a > frame." > + self assert: needsFrame. > + needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > + "Note that ReceiverResultReg remains live after both > + ceStoreContextInstVarTrampoline and ceStoreCheckTrampoline." > + self ensureReceiverResultRegContainsSelf. > + self ssPop: 1. > + self ssAllocateCallReg: ClassReg and: SendNumArgsReg. "for > ceStoreContextInstVarTrampoline" > + self ssPush: 1. > + objectRepresentation > + genLoadSlot: SenderIndex > + sourceReg: ReceiverResultReg > + destReg: TempReg. > + self ssStoreAndReplacePop: popBoolean toReg: ClassReg. > + "stack is flushed except maybe ssTop if popBoolean is false. > + ssTop is a SSregister in this case due to #ssStoreAndReplacePop: > + to avoid a second indirect read / annotation in case of > SSConstant > + or SSBaseRegister" > + self ssFlushTo: simStackPtr. > + jmpSingle := objectRepresentation > genJumpNotSmallIntegerInScratchReg: TempReg. > + self MoveCq: slotIndex R: SendNumArgsReg. > + self CallRT: ceStoreContextInstVarTrampoline. > + jmpDone := self Jump: 0. > + jmpSingle jmpTarget: self Label. > + objectRepresentation > + genStoreWithImmutabilityCheckSourceReg: ClassReg > + slotIndex: slotIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > + needsStoreCheck: needStoreCheck > + needRestoreRcvr: true. > + jmpDone jmpTarget: self Label. > + ^0! > > Item was added: > + ----- Method: > StackToRegisterMappingCogit>>genImmutabilityCheckStorePop:ReceiverVariable: > (in category 'bytecode generator support') ----- > + genImmutabilityCheckStorePop: popBoolean ReceiverVariable: slotIndex > + <inline: true> > + | needStoreCheck | > + self assert: needsFrame. > + needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > + "Note that ReceiverResultReg remains live after the trampoline." > + self ensureReceiverResultRegContainsSelf. > + self ssAllocateRequiredReg: ClassReg. > + self ssStoreAndReplacePop: popBoolean toReg: ClassReg. > + self ssFlushTo: simStackPtr. > + objectRepresentation > + genStoreWithImmutabilityCheckSourceReg: ClassReg > + slotIndex: slotIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > + needsStoreCheck: needStoreCheck > + needRestoreRcvr: true. > + > + ^ 0! > > Item was changed: > ----- Method: StackToRegisterMappingCogit>>genStorePop:LiteralVariable: > (in category 'bytecode generator support') ----- > genStorePop: popBoolean LiteralVariable: litVarIndex > <inline: false> > - | topReg association needStoreCheck immutabilityFailure | > - "The only reason we assert needsFrame here is that in a frameless > method > - ReceiverResultReg must and does contain only self, but the > ceStoreCheck > - trampoline expects the target of the store to be in > ReceiverResultReg. So > - in a frameless method we would have a conflict between the > receiver and > - the literal store, unless we we smart enough to realise that > ReceiverResultReg > - was unused after the literal variable store, unlikely given that > methods > - return self by default." > - self assert: needsFrame. > - self cppIf: IMMUTABILITY ifTrue: [ self ssFlushTo: simStackPtr - 1 > ]. > - "N.B. No need to check the stack for references because we > generate code for > - literal variable loads that stores the result in a register, > deferring only the register push." > - needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > - association := self getLiteral: litVarIndex. > - optStatus isReceiverResultRegLive: false. > - self ssAllocateRequiredReg: ReceiverResultReg. "for ceStoreCheck > call in genStoreSourceReg: has to be ReceiverResultReg" > - self genMoveConstant: association R: ReceiverResultReg. > - objectRepresentation genEnsureObjInRegNotForwarded: > ReceiverResultReg scratchReg: TempReg. > self > cppIf: IMMUTABILITY > + ifTrue: [ ^ self genImmutabilityCheckStorePop: popBoolean > LiteralVariable: litVarIndex ] > + ifFalse: [ ^ self genVanillaStorePop: popBoolean > LiteralVariable: litVarIndex ] > + ! > - ifTrue: > - [ self ssAllocateRequiredReg: ClassReg. > - topReg := ClassReg. > - self ssStoreAndReplacePop: popBoolean toReg: > ClassReg. > - "stack is flushed except maybe ssTop if > popBoolean is false. > - ssTop is a SSregister in this case due to > #ssStoreAndReplacePop: > - to avoid a second indirect read / annotation in > case of SSConstant > - or SSBaseRegister" > - self ssFlushTo: simStackPtr. > - immutabilityFailure := objectRepresentation > - > genImmutableCheck: ReceiverResultReg > - > slotIndex: ValueIndex > - > sourceReg: ClassReg > - > scratchReg: TempReg > - > needRestoreRcvr: false ] > - ifFalse: > - [ topReg := self allocateRegForStackEntryAt: 0 > notConflictingWith: (self registerMaskFor: ReceiverResultReg). > - self ssStorePop: popBoolean toReg: topReg ]. > - traceStores > 0 ifTrue: > - [self MoveR: topReg R: TempReg. > - self CallRT: ceTraceStoreTrampoline]. > - objectRepresentation > - genStoreSourceReg: topReg > - slotIndex: ValueIndex > - destReg: ReceiverResultReg > - scratchReg: TempReg > - inFrame: needsFrame > - needsStoreCheck: needStoreCheck. > - self cppIf: IMMUTABILITY ifTrue: [ immutabilityFailure jmpTarget: > self Label ]. > - ^ 0! > > Item was changed: > ----- Method: > StackToRegisterMappingCogit>>genStorePop:MaybeContextReceiverVariable: (in > category 'bytecode generator support') ----- > genStorePop: popBoolean MaybeContextReceiverVariable: slotIndex > <inline: false> > - | jmpSingle jmpDone needStoreCheck immutabilityFailure | > - <var: #jmpSingle type: #'AbstractInstruction *'> > - <var: #jmpDone type: #'AbstractInstruction *'> > - "The reason we need a frame here is that assigning to an inst var > of a context may > - involve wholesale reorganization of stack pages, and the only way > to preserve the > - execution state of an activation in that case is if it has a > frame." > - self assert: needsFrame. > - self cppIf: IMMUTABILITY ifTrue: [ self ssFlushTo: simStackPtr - 1 > ]. > - self ssFlushUpThroughReceiverVariable: slotIndex. > - needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > - "Note that ReceiverResultReg remains live after both > - ceStoreContextInstVarTrampoline and ceStoreCheckTrampoline." > - self ensureReceiverResultRegContainsSelf. > - self ssPop: 1. > - self ssAllocateCallReg: ClassReg and: SendNumArgsReg. "for > ceStoreContextInstVarTrampoline" > - self ssPush: 1. > - objectRepresentation > - genLoadSlot: SenderIndex > - sourceReg: ReceiverResultReg > - destReg: TempReg. > self > cppIf: IMMUTABILITY > + ifTrue: [ ^ self genImmutabilityCheckStorePop: popBoolean > MaybeContextReceiverVariable: slotIndex ] > + ifFalse: [ ^ self genVanillaStorePop: popBoolean > MaybeContextReceiverVariable: slotIndex ]! > - ifTrue: > - [ self ssStoreAndReplacePop: popBoolean toReg: > ClassReg. > - "stack is flushed except maybe ssTop if > popBoolean is false. > - ssTop is a SSregister in this case due to > #ssStoreAndReplacePop: > - to avoid a second indirect read / annotation in > case of SSConstant > - or SSBaseRegister" > - self ssFlushTo: simStackPtr. ] > - ifFalse: [ self ssStorePop: popBoolean toReg: ClassReg ]. > - jmpSingle := objectRepresentation > genJumpNotSmallIntegerInScratchReg: TempReg. > - self MoveCq: slotIndex R: SendNumArgsReg. > - self CallRT: ceStoreContextInstVarTrampoline. > - jmpDone := self Jump: 0. > - jmpSingle jmpTarget: self Label. > - traceStores > 0 ifTrue: > - [self MoveR: ClassReg R: TempReg. > - self CallRT: ceTraceStoreTrampoline]. > - self > - cppIf: IMMUTABILITY > - ifTrue: > - [ immutabilityFailure := objectRepresentation > - > genImmutableCheck: ReceiverResultReg > - > slotIndex: ValueIndex > - > sourceReg: ClassReg > - > scratchReg: TempReg > - > needRestoreRcvr: true ]. > - objectRepresentation > - genStoreSourceReg: ClassReg > - slotIndex: slotIndex > - destReg: ReceiverResultReg > - scratchReg: TempReg > - inFrame: true > - needsStoreCheck: needStoreCheck. > - jmpDone jmpTarget: self Label. > - self cppIf: IMMUTABILITY ifTrue: [ immutabilityFailure jmpTarget: > self Label ]. > - ^0! > > Item was changed: > ----- Method: StackToRegisterMappingCogit>>genStorePop:ReceiverVariable: > (in category 'bytecode generator support') ----- > genStorePop: popBoolean ReceiverVariable: slotIndex > <inline: false> > - | topReg needStoreCheck immutabilityFailure | > - self cppIf: IMMUTABILITY ifTrue: [ self assert: needsFrame. self > ssFlushTo: simStackPtr - 1 ]. > - self ssFlushUpThroughReceiverVariable: slotIndex. > - needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > - "Note that ReceiverResultReg remains live after > ceStoreCheckTrampoline." > - self ensureReceiverResultRegContainsSelf. > self > cppIf: IMMUTABILITY > + ifTrue: [ ^ self genImmutabilityCheckStorePop: popBoolean > ReceiverVariable: slotIndex ] > + ifFalse: [ ^ self genVanillaStorePop: popBoolean > ReceiverVariable: slotIndex ] > + ! > - ifTrue: > - [ self ssAllocateRequiredReg: ClassReg. > - topReg := ClassReg. > - self ssStoreAndReplacePop: popBoolean toReg: > ClassReg. > - "stack is flushed except maybe ssTop if > popBoolean is false. > - ssTop is a SSregister in this case due to > #ssStoreAndReplacePop: > - to avoid a second indirect read / annotation in > case of SSConstant > - or SSBaseRegister" > - self ssFlushTo: simStackPtr. > - immutabilityFailure := objectRepresentation > - > genImmutableCheck: ReceiverResultReg > - > slotIndex: slotIndex > - > sourceReg: ClassReg > - > scratchReg: TempReg > - > needRestoreRcvr: true ] > - ifFalse: > - [ topReg := self allocateRegForStackEntryAt: 0 > notConflictingWith: (self registerMaskFor: ReceiverResultReg). > - self ssStorePop: popBoolean toReg: topReg ]. > - traceStores > 0 ifTrue: > - [ self MoveR: topReg R: TempReg. > - self evaluateTrampolineCallBlock: [ self CallRT: > ceTraceStoreTrampoline ] protectLinkRegIfNot: needsFrame ]. > - objectRepresentation > - genStoreSourceReg: topReg > - slotIndex: slotIndex > - destReg: ReceiverResultReg > - scratchReg: TempReg > - inFrame: needsFrame > - needsStoreCheck: needStoreCheck. > - self cppIf: IMMUTABILITY ifTrue: [ immutabilityFailure jmpTarget: > self Label ]. > - ^ 0! > > Item was changed: > ----- Method: StackToRegisterMappingCogit>>genStorePop:RemoteTemp:At: > (in category 'bytecode generator support') ----- > genStorePop: popBoolean RemoteTemp: slotIndex At: remoteTempIndex > <inline: false> > | topReg needStoreCheck | > "The only reason we assert needsFrame here is that in a frameless > method > ReceiverResultReg must and does contain only self, but the > ceStoreCheck > trampoline expects the target of the store to be in > ReceiverResultReg. So > in a frameless method we would have a conflict between the > receiver and > the temote temp store, unless we we smart enough to realise that > ReceiverResultReg was unused after the literal variable store, > unlikely given > that methods return self by default." > self assert: needsFrame. > "N.B. No need to check the stack for references because we > generate code for > remote temp loads that stores the result in a register, deferring > only the register push." > needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > topReg := self allocateRegForStackEntryAt: 0 notConflictingWith: > (self registerMaskFor: ReceiverResultReg). > self ssAllocateRequiredReg: ReceiverResultReg. > optStatus isReceiverResultRegLive: false. > self ssStoreAndReplacePop: popBoolean toReg: topReg. > self MoveMw: (self frameOffsetOfTemporary: remoteTempIndex) r: > FPReg R: ReceiverResultReg. > - traceStores > 0 ifTrue: > - [ self MoveR: topReg R: TempReg. > - self CallRT: ceTraceStoreTrampoline. ]. > ^objectRepresentation > genStoreSourceReg: topReg > slotIndex: slotIndex > destReg: ReceiverResultReg > scratchReg: TempReg > inFrame: needsFrame > needsStoreCheck: needStoreCheck! > > Item was added: > + ----- Method: StackToRegisterMappingCogit>>genTraceStores (in category > 'bytecode generator support') ----- > + genTraceStores > + <inline: true> > + traceStores > 0 ifTrue: > + [ self MoveR: ClassReg R: TempReg. > + self CallRT: ceTraceStoreTrampoline ].! > > Item was added: > + ----- Method: > StackToRegisterMappingCogit>>genVanillaStorePop:LiteralVariable: (in > category 'bytecode generator support') ----- > + genVanillaStorePop: popBoolean LiteralVariable: litVarIndex > + <inline: true> > + | topReg association needStoreCheck | > + "The only reason we assert needsFrame here is that in a frameless > method > + ReceiverResultReg must and does contain only self, but the > ceStoreCheck > + trampoline expects the target of the store to be in > ReceiverResultReg. So > + in a frameless method we would have a conflict between the > receiver and > + the literal store, unless we we smart enough to realise that > ReceiverResultReg > + was unused after the literal variable store, unlikely given that > methods > + return self by default." > + self assert: needsFrame. > + "N.B. No need to check the stack for references because we > generate code for > + literal variable loads that stores the result in a register, > deferring only the register push." > + needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > + association := self getLiteral: litVarIndex. > + optStatus isReceiverResultRegLive: false. > + self ssAllocateRequiredReg: ReceiverResultReg. "for ceStoreCheck > call in genStoreSourceReg: has to be ReceiverResultReg" > + self genMoveConstant: association R: ReceiverResultReg. > + objectRepresentation genEnsureObjInRegNotForwarded: > ReceiverResultReg scratchReg: TempReg. > + topReg := self allocateRegForStackEntryAt: 0 notConflictingWith: > (self registerMaskFor: ReceiverResultReg). > + self ssStorePop: popBoolean toReg: topReg. > + objectRepresentation > + genStoreSourceReg: topReg > + slotIndex: ValueIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > + inFrame: needsFrame > + needsStoreCheck: needStoreCheck. > + ^ 0! > > Item was added: > + ----- Method: > StackToRegisterMappingCogit>>genVanillaStorePop:MaybeContextReceiverVariable: > (in category 'bytecode generator support') ----- > + genVanillaStorePop: popBoolean MaybeContextReceiverVariable: slotIndex > + <inline: true> > + | jmpSingle jmpDone needStoreCheck | > + <var: #jmpSingle type: #'AbstractInstruction *'> > + <var: #jmpDone type: #'AbstractInstruction *'> > + "The reason we need a frame here is that assigning to an inst var > of a context may > + involve wholesale reorganization of stack pages, and the only way > to preserve the > + execution state of an activation in that case is if it has a > frame." > + self assert: needsFrame. > + self ssFlushUpThroughReceiverVariable: slotIndex. > + needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > + "Note that ReceiverResultReg remains live after both > + ceStoreContextInstVarTrampoline and ceStoreCheckTrampoline." > + self ensureReceiverResultRegContainsSelf. > + self ssPop: 1. > + self ssAllocateCallReg: ClassReg and: SendNumArgsReg. "for > ceStoreContextInstVarTrampoline" > + self ssPush: 1. > + objectRepresentation > + genLoadSlot: SenderIndex > + sourceReg: ReceiverResultReg > + destReg: TempReg. > + self ssStorePop: popBoolean toReg: ClassReg. > + jmpSingle := objectRepresentation > genJumpNotSmallIntegerInScratchReg: TempReg. > + self MoveCq: slotIndex R: SendNumArgsReg. > + self CallRT: ceStoreContextInstVarTrampoline. > + jmpDone := self Jump: 0. > + jmpSingle jmpTarget: self Label. > + objectRepresentation > + genStoreSourceReg: ClassReg > + slotIndex: slotIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > + inFrame: true > + needsStoreCheck: needStoreCheck. > + jmpDone jmpTarget: self Label. > + > + ^0! > > Item was added: > + ----- Method: > StackToRegisterMappingCogit>>genVanillaStorePop:ReceiverVariable: (in > category 'bytecode generator support') ----- > + genVanillaStorePop: popBoolean ReceiverVariable: slotIndex > + <inline: true> > + | topReg needStoreCheck | > + self ssFlushUpThroughReceiverVariable: slotIndex. > + needStoreCheck := (objectRepresentation isUnannotatableConstant: > self ssTop) not. > + "Note that ReceiverResultReg remains live after > ceStoreCheckTrampoline." > + self ensureReceiverResultRegContainsSelf. > + topReg := self allocateRegForStackEntryAt: 0 notConflictingWith: > (self registerMaskFor: ReceiverResultReg). > + self ssStorePop: popBoolean toReg: topReg. > + objectRepresentation > + genStoreSourceReg: topReg > + slotIndex: slotIndex > + destReg: ReceiverResultReg > + scratchReg: TempReg > + inFrame: needsFrame > + needsStoreCheck: needStoreCheck. > + ^ 0! > >

2 4

vm problem on cog an stack v3
by Nicolas Cellier 31 Mar '16

31 Mar '16

Hi, before I continue, i've noticed that the large integer multiply seems broken on v3 object memory (cog & stack) Note that this does not happen on Spur. This is independent of my recent changes of LargeIntegers plugin as it happens BEFORE these changes and is related to primitive 29 rather than to the plugin... Here are the symptoms: halfPower := 10000. s := 111111111111111. head := s quo: halfPower. tail := s - (head * halfPower). { head as: ByteArray. (1 to: halfPower digitLength) collect: [:i | halfPower digitAt: i] as: ByteArray. (head*halfPower) as: ByteArray. }. the correct result is: #(#[199 25 70 150 2] #[16 39] #[112 237 78 18 14 101]) the wrong result I obtained with SVN revision 3651 compiled by myself is: #(#[199 25 70 150 2] #[16 39] #[112 237 78 18 254 61]) The most significant bits (above 32) are wrong... The pattern I obtain is (with most significant bit put back left) 2r00111101 << 8 + 2r11111110 "wrong result" 2r01100101 << 8 + 2r00001110 "Correct result" I completely fail to infer what's going on from this pattern... This is on MacOSX clang --version Apple LLVM version 7.3.0 (clang-703.0.29) Target: x86_64-apple-darwin15.4.0 This goes thru primitiveMultiplyLargeIntegers (29) oopResult := self magnitude64BitIntegerFor: result neg: aIsNegative ~= bIsNegative. -> sz > 4 ifTrue: [objectMemory storeLong64: 0 ofObject: newLargeInteger withValue: magnitude] (which I changed recently) then: storeLong64: longIndex ofObject: oop withValue: value <var: #value type: #sqLong> self flag: #endianness. self long32At: oop + self baseHeaderSize + (longIndex << 3) put: (self cCode: [value] inSmalltalk: [value bitAnd: 16rFFFFFFFF]); long32At: oop + self baseHeaderSize + (longIndex << 3) + 4 put: (value >> 32). ^value I don't see anything wrong with this code... Well, using a shift on signed value is not that good, but it works for at least 3 reasons: - we throw the signBit extension away - slang inlining misses the signedness difference, and the generated C code is correct. - Anyway, in our case, the sign bit was 0... Previous implementation in magnitude64BitIntegerFor:neg: was: sz > 4 ifTrue: [objectMemory storeLong32: 1 ofObject: newLargeInteger withValue: magnitude >> 32]. objectMemory storeLong32: 0 ofObject: newLargeInteger withValue: (self cCode: [magnitude] inSmalltalk: [magnitude bitAnd: 16rFFFFFFFF]) Not much different, except that the high 32 bits and low 32 bits are written in a different order... If I had a server I'd like to bisect - from which version does this happens - for which OS - for which compiler Without such information, I think I'll have to debug it either thru simulator or directly in gdb, but I feel like I'm really losing my time :( And I've got a 2nd problem like this one...

4 9

64-bit WideString access (was [squeak-dev] New Cog VMs available)
by Eliot Miranda 31 Mar '16

31 Mar '16

Hi Levente, On Tue, Mar 29, 2016 at 7:10 PM, Levente Uzonyi <leves(a)caesar.elte.hu> wrote: > Hi Eliot, > > I think something went wrong with those semaphores. Connecting to a Socket > seems to wait for the full 45 seconds timeout with this VM. To reproduce, > just try opening any remote MC repository. This happens with > cogspur64linuxht. I haven't tried any other versions yet. > Not sure what's going on here but I can reproduce the bug. > > There's another issue with WideStrings where, #at: (primitive 63) fails > for a few instances. This issue is present in 3643 as well. To reproduce > it, open the Monticello Browser, select the Multilingual package, select > the Trunk repository and press Changes. You'll see that two methods are > reported to have changed (even though they haven't). Selecting any of them > will pop up a debugger with the primitive failure. The common symptom is > that #basicAt: will work for the index, for which #at: fails, and it will > return the number 16rD80000FF. So this is interesting. It may boil down to leading character issues. First, the bug is that the JIT and the Interpreter don't agree on the semantics of WideString>>#at: when the value at the index is "out of range". Notionally the Spur VM restricts characters to the range 0 through 16r3FFFFFFF, to keep them in the +ve SmallInteger range. The JIT primitive fails if asked to answer a Character outside that range, and 16rD80000FF is outside, hence the JIT primitive fails. The Interpreter primitive does not (yet) fail. There's an optimization in the JIT that if a primitive deals with all possible inputs and the method is written not to use an error code then the primitive will fail without invoking the interpreter primitive. This a) means the JIT doesn't have to set error codes correctly and b) unless the programmer wants the failure code failure is a bit quicker. But in this case I got caught. Because the Interpreter primitive does not fail, changing the method to include the error code the JIT now calls the Interpreter primitive when the JIT primitive fails, the interpreter primitive does not fail, and the character is answered anyway. So there are a few things going on here... a) in loading the Monticello snapshot for Multilingual-ul.210 a WideString gets created for JapaneseEnvironment class>>#flapTabTextFor:in: that has 16rD80000FF at 213 and 275 in the source, whereas (JapaneseEnvironment class>>#flapTabTextFor:in:) getSource string contains Character value: 27 (Character escape) at 213 and 275, and Character value: 16r14030FC at 214 and 276, but weirdly the strings have the same size, and are equal everywhere else. So somewhere in the Monticello package read (or write?) the escape is being rewritten to Character value: 16rD80000FF. b) I'm confused^H^H^H^H^H^H^H^Hnot sure what to do about the range of acceptable characters. It seems right to me to only use 30 bits for characters; that's more than enough for Unicode and it seems arbitrary to allow e.g. 61 bit Characters in 64-bits and 30 bit Characters in 32-bit (makes sense with integers because of LargeInteger). So I should either make WideString>>at:[put:] fail for out of range characters in all versions of the primitives, and modify the methods so they yield a sensible error message, or should I make the JIT primitive not fail? I want to fail for out-of-range characters, but not until we understand the Monticello issue that maps Character escape to Character value: 16rD80000FF. If you want to experiment with the system then here are some variants on WideString>>at:[put:] that will be useful (also attached). To see the JIT primitives fail, but get a sensible error message, use WideString>>at: index "Answer the Character stored in the field of the receiver indexed by the argument. Primitive. Fail if the index argument is not an Integer or is out of bounds. Essential. See Object documentation whatIsAPrimitive." <primitive: 63> ^index isInteger ifTrue: [(index >= 1 and: [index <= self size]) ifTrue: [self error: 'illegal character value ', ((self basicAt: index) storeStringBase: 16)] ifFalse: [self errorSubscriptBounds: index]] ifFalse: [index isNumber ifTrue: [self at: index asInteger] ifFalse: [self errorNonIntegerIndex]] WideString>>at: index put: aCharacter "Store the Character into the field of the receiver indicated by the index. Primitive. Fail if the index is not an Integer or is out of bounds, or if the argument is not a Character. Essential. See Object documentation whatIsAPrimitive." <primitive: 64> ^aCharacter isCharacter ifTrue: [index isInteger ifTrue: [(index >= 1 and: [index <= self size]) ifTrue: [self error: 'illegal character value ', (aCharacter asInteger storeStringBase: 16)] ifFalse: [self errorSubscriptBounds: index]] ifFalse: [self errorNonIntegerIndex]] ifFalse: [self errorImproperStore] To suppress the errors and have the interpreter primitives back up (and answer out-of-range Characters) simply include an error code: at: index "Answer the Character stored in the field of the receiver indexed by the argument. Primitive. Fail if the index argument is not an Integer or is out of bounds. Essential. See Object documentation whatIsAPrimitive." <primitive: 63 error: ec> ^index isInteger ifTrue: [self errorSubscriptBounds: index] ifFalse: [index isNumber ifTrue: [self at: index asInteger] ifFalse: [self errorNonIntegerIndex]] Again, the decision for the VM is whether to drop the pretence of Characters being 30 bits (not something I'm happy with) or whether to make sure the interpreter WideString primitives also fail for out-of-range characters, and for us to fix Monticello snapshot read (write?) for the Character escape, Character value: 16r14030FC sequence. The problem is that there are ways of creating strings that bypass both Character class>>value: and WideString>>at:put:. Compare these: This produces a WideString full of illegal characters via adoptInstance: | s | s := Bitmap new: 16 withAll: 16rD80000FF. WideString adoptInstance: s. s. This does the same via basicAt:[put:] which treat WideString as arrays of 32-bit unsigned integers (which is what they are in memory). | s | s := WideString new: 16. 1 to: s size do: [:i| s basicAt: i put: 16rD80000FF] This fails to do so because Character class>>#value: has a range check WideString new: 16 withAll: (Character value: 16rD80000FF) This fails if the JIT primitive is in use: (tryPrimitive:withArgs: side-steps the JIT primitive, calls the interpreter primitive which does answer the out-of-range character) | illegalCharacter s | illegalCharacter := ((WideString new: 1) basicAt: 1 put: 16rD80000FF; tryPrimitive: 63 withArgs: #(1)). s := WideString new: 16. 1 to: s size do: [:i| s at: i put: illegalCharacter] I'm open to being called idiotic here (please weigh in) but I think the "right thing" is a) make the Interpreter wide string primitives also fail for character access outside of 30 bits. b) change the error handling on WideString>>at:[put:] to give sensible error messages when attempting character access outside of 30 bits. c) fix the character encoding issue Here's the source of the method: !JapaneseEnvironment class methodsFor: 'language methods' stamp: 'yo 8/3/2004 21 :25'! flapTabTextFor: aString in: aFlapTab | string | string := super flapTabTextFor: aString. string isEmptyOrNil ifTrue: [^ self]. string := aFlapTab orientation == #vertical ifTrue: [string copyReplaceAll: '<E3><83><BC>' with: '<EF><BD><9C>'] ifFalse: [string copyReplaceAll: '<EF><BD><9C>' with: '<E3><83><BC>']. ^ string. !]lang[(213 1 9 1 41 1 9 1 16)0,5,0,5,0,5,0,5,0! ! This is read differently in the CompiledMethod>>getSource case to the Monticello snapshot load case. Why? Here's a model of that method source: | w | w := WideString with: 102 asCharacter with: 108 asCharacter with: 27 asCharacter with: 16r14030FC asCharacter. ByteString streamContents: [:s| s nextChunkPut: w] 'flー!]lang[(3 1)0,5!' This doit implies that it's not the converters that are at fault: | w | w := WideString with: 102 asCharacter with: 108 asCharacter with: 27 asCharacter with: 16r14030FC asCharacter. (TextConverter allSubclasses reject: [:tcc| tcc includesBehavior: ByteTextConverter]) reject: [:tcc| | r | r := (ByteString streamContents: [:s| s nextChunkPut: w]) readStream. (tcc new parseLangTagFor: (r next: 4) fromStream: r) = w] an OrderedCollection() i.e. all the multibyte converters answer the string unaltered. So where is this 16rD80000FF bit pattern comming from? P.S. Digging further I see it /shouldn't/ be a leading character issue. There's just enough room for a 22 bit Unicode character code and an 8-bit leading character language code: Character>>encodedCharSet self asInteger < 16r400000 ifTrue: [ ^Unicode ]. "Shortcut" ^EncodedCharSet charsetAt: self leadingChar Character>>leadingChar "Answer the value of the 8 highest bits which is used to identify the language. This is mostly used for east asian languages CJKV as a workaround against unicode han-unification." ^ self asInteger bitShift: -22 16rFF << 22 = 16r3F,C0,00,00 Levente > > > On Tue, 29 Mar 2016, Eliot Miranda wrote: > > ... at http://www.mirandabanda.org/files/Cog/VM/VM.r3663 >> >> >> CogVM binaries as per VMMaker.oscog-eem.1746/r3663 >> >> General: >> >> Fix loss of signals to e.g. a socket's readSemaphore when data available: >> Fix a bug in sqAtomicOps.h where the assumption that the intrinsic atomic >> add >> operation will apply to 16-bit variables is false on clang. So on x86 >> and x64 >> use inline assembly with gcc and clang since this is known to work. >> Improve the doSignalSemaphores code in sqExternalSemaphores.c so that the >> tide >> variables are initialized at the right point and so there's only one copy >> of >> the signalling code. >> >> Allow primitiveUtcWithOffset to accept an optional parameter with an >> array or object with two or more slots to store UTC posix microseconds >> and time zone offset in seconds. Compatibility with VMM trunk. >> >> Add a Smalltalk epoch version of it, primitiveUtcAndTimezoneOffset, >> and give it primitive #244. >> >> Fix signed/unsigned arithmetic issues in 64-bit microsecond clock code. >> This fixes the bug whereby >> Time localMicrosecondClock - Time utcMicrosecondClock // 1000000 >> would answer something with an odd second, a multiple of 3600 plus 1. >> >> General Integer conversion routines: >> Several clean ups to integer conversion routines. >> >> Simplify bit operations using positiveMachineIntegerValueOf:/ >> positiveMachineIntegerFor: rather than doing 32/64 bits dissertation. >> >> >> Spur Cogit: >> Rewrote identity primitive to check for forwarders only when the identity >> of >> objects is different. The previous version would crash if the argument >> was a >> forwarder to an immediate. >> >> Spur VMs: >> Now that UUIDs are created with purely image-based code in Squeak trunk, >> make >> the UUIDPlugin external in all Squeak Spur VMs. >> >> >> Plugins: >> Make FFI load symbol fail after (when find function fails). >> In Pharo, we allow to load global functions (so we can control world >> windows >> through FFI). >> >> LargeIntegersPlugin: >> >> Finish 1st round of LargeIntegers refactoring >> >> - simplify the left and right shift >> - use as much unsigned arithmetic as possible >> - homogenize type declaration within the plugin >> - remove unused digitOf:at: >> >> >> Slang: >> Beware: sending abs to an unsigned int will re-interpret the int as >> signed and >> will generate C compiler warnings. The new version uses SQABS and SQLABS >> macros >> on sqInt and sqLong vars, and fabs and fabsf on double and float args, >> respectively, and llabs on long long, __int64 vars. >> >> >> Mac OS X: >> Unswap the Squeak and Newspeak localized strings >> >> _,,,^..^,,,_ >> best, Eliot >> > _,,,^..^,,,_ best, Eliot

2 1

VM Maker: MemoryAccess-dtl.6.mcz
by commits＠source.squeak.org 31 Mar '16

31 Mar '16

David T. Lewis uploaded a new version of MemoryAccess to project VM Maker: http://source.squeak.org/VMMaker/MemoryAccess-dtl.6.mcz ==================== Summary ==================== Name: MemoryAccess-dtl.6 Author: dtl Time: 30 March 2016, 7:28:23.155 pm UUID: 6fabecce-e2aa-4bec-8b2e-f421802beda7 Ancestors: MemoryAccess-dtl.5 Convert to pragma declarations using Pragmatizer =============== Diff against MemoryAccess-dtl.5 =============== Item was changed: ----- Method: MemoryAccess>>byteAt: (in category 'object memory access') ----- byteAt: oop "Answer the byte at an object memory location. The result is a signed sqInt value in the range 0 through 255." "sqInt byteAt(sqInt oop) { return byteAtPointer(pointerForOop(oop)); }" + <inline: true> - self inline: true. ^ self byteAtPointer: (self pointerForOop: oop)! Item was changed: ----- Method: MemoryAccess>>byteAt:put: (in category 'object memory access') ----- byteAt: oop put: val "Cast a sqInt value to signed integer and store its low order byte at the given object memory location." "sqInt byteAtput(sqInt oop, int val) { return byteAtPointerput(pointerForOop(oop), val); }" + <inline: true> + <var: #val type: 'int'> - self inline: true. - self var: #val type: 'int'. ^ self byteAtPointer: (self pointerForOop: oop) put: val! Item was changed: ----- Method: MemoryAccess>>byteAtPointer: (in category 'map address to value') ----- byteAtPointer: ptr "Answer the unsigned byte value at a machine address. The result is a signed sqInt value in the range 0 through 255." "sqInt byteAtPointer(char *ptr) { return (sqInt)(*((unsigned char *)ptr)) }" + <inline: true> + <var: #ptr type: 'char *'> - self inline: true. - self var: #ptr type: 'char *'. ^ self cCoerce: ((self cCoerce: ptr to: 'unsigned char *') at: 0) to: 'sqInt'! Item was changed: ----- Method: MemoryAccess>>byteAtPointer:put: (in category 'map address to value') ----- byteAtPointer: ptr put: val "Cast a sqInt value to signed integer and store its low order byte at the given machine address." "sqInt byteAtPointerput(char *ptr, int val) { return (sqInt)(*((unsigned char *)ptr)= (unsigned char)val) }" + <inline: true> + <var: #ptr type: 'char *'> + <var: #val type: 'int'> - self inline: true. - self var: #ptr type: 'char *'. - self var: #val type: 'int'. ^ self cCoerce: ((self cCoerce: ptr to: 'unsigned char *') at: 0 put: val) to: 'sqInt'! Item was changed: ----- Method: MemoryAccess>>intAt: (in category 'object memory access') ----- intAt: oop "Answer the unsigned integer value at an object memory location. The result is a signed sqInt value with binary value in the range 0 through 16rFFFFFFFF. If the size of sqInt is 8 bytes (64-bit object memory) and size of integer is 4, then the high order 4 bytes of the result are zero." "sqInt intAt(sqInt oop) { return intAtPointer(pointerForOop(oop)); }" + <inline: true> - self inline: true. ^ self intAtPointer: (self pointerForOop: oop)! Item was changed: ----- Method: MemoryAccess>>intAt:put: (in category 'object memory access') ----- intAt: oop put: val "Cast a sqInt value to unsigned integer and store its low order bytes at the given object memory location. Assuming that size of integer is 4, then four bytes are stored regardless of the size of sqInt (32-bit or 64-bit object memory)." "sqInt intAtput(sqInt oop, int val) { return intAtPointerput(pointerForOop(oop), val); }" + <inline: true> + <var: #val type: 'int'> - self inline: true. - self var: #val type: 'int'. ^ self intAtPointer: (self pointerForOop: oop) put: val! Item was changed: ----- Method: MemoryAccess>>intAtPointer: (in category 'map address to value') ----- intAtPointer: ptr "Answer the unsigned integer value at a machine address. The result is a signed sqInt value with binary value in the range 0 through 16rFFFFFFFF. If the size of sqInt is 8 bytes (64-bit object memory) and size of integer is 4, then the high order 4 bytes of the result are zero." "sqInt intAtPointer(char *ptr) { return (sqInt)(*((unsigned int *)ptr)) }" + <inline: true> + <var: #ptr type: 'char *'> - self inline: true. - self var: #ptr type: 'char *'. ^ self cCoerce: ((self cCoerce: ptr to: 'unsigned int *') at: 0) to: 'sqInt'! Item was changed: ----- Method: MemoryAccess>>intAtPointer:put: (in category 'map address to value') ----- intAtPointer: ptr put: val "Cast a sqInt value to unsigned integer and store its low order bytes at the given machine address. Assuming that size of integer is 4, then four bytes are stored regardless of the size of sqInt (32-bit or 64-bit object memory)." "sqInt intAtPointerput(char *ptr, int val) { return (sqInt)(*((unsigned int *)ptr)= (int)val) }" + <inline: true> + <var: #ptr type: 'char *'> + <var: #val type: 'unsigned int'> - self inline: true. - self var: #ptr type: 'char *'. - self var: #val type: 'unsigned int'. ^ self cCoerce: ((self cCoerce: ptr to: 'unsigned int *') at: 0 put: val) to: 'sqInt'! Item was changed: ----- Method: MemoryAccess>>long32At: (in category 'object memory access') ----- long32At: ptr "#define long32At intAt" + <inline: true> - self inline: true. ^ self intAt: ptr ! Item was changed: ----- Method: MemoryAccess>>long32At:put: (in category 'object memory access') ----- long32At: ptr put: val "#define long32Atput intAtput" + <inline: true> + <var: #val type: 'int'> - self inline: true. - self var: #val type: 'int'. ^ self intAt: ptr put: val ! Item was changed: ----- Method: MemoryAccess>>longAt: (in category 'object memory access') ----- longAt: oop "Answer the object memory word, an sqInt, located at an object memory location. In the original 32-bit object memory implemention, word size was equivalent to C type long. For 64-bit object memory, the method name may be misleading because the sqInt value is of C type long long." "sqInt longAt(sqInt oop) { return longAtPointer(pointerForOop(oop)); }" + <inline: true> - self inline: true. ^ self longAtPointer: (self pointerForOop: oop) ! Item was changed: ----- Method: MemoryAccess>>longAt:put: (in category 'object memory access') ----- longAt: oop put: val "Store an object memory word, an sqInt, at the given object memory location. In the original 32-bit object memory implemention, word size was equivalent to C type long. For 64-bit object memory, the method name may be misleading because the sqInt value is of C type long long." "sqInt longAtput(sqInt oop, sqInt val) { return longAtPointerput(pointerForOop(oop), val); }" + <inline: true> - self inline: true. ^ self longAtPointer: (self pointerForOop: oop) put: val ! Item was changed: ----- Method: MemoryAccess>>longAtPointer: (in category 'map address to value') ----- longAtPointer: ptr "Answer the object memory word, an sqInt, located at a machine address. In the original 32-bit object memory implemention, word size was equivalent to C type long. For 64-bit object memory, the method name may be misleading because the sqInt value is of C type long long." "sqInt longAtPointer(char *ptr) { return (sqInt)(*((sqInt *)ptr)) }" + <inline: true> + <var: #ptr type: 'char *'> - self inline: true. - self var: #ptr type: 'char *'. ^ self cCoerce: ((self cCoerce: ptr to: 'sqInt *') at: 0) to: 'sqInt' ! Item was changed: ----- Method: MemoryAccess>>longAtPointer:put: (in category 'map address to value') ----- longAtPointer: ptr put: val "Store an object memory word, an sqInt, at the given machine address. In the original 32-bit object memory implemention, word size was equivalent to C type long. For 64-bit object memory, the method name may be misleading because the sqInt value is of C type long long." "sqInt longAtPointerput(char *ptr, sqInt val) { return (sqInt)(*((sqInt *)ptr)= (sqInt)val) }" + <inline: true> + <var: #ptr type: 'char *'> - self inline: true. - self var: #ptr type: 'char *'. ^ self cCoerce: ((self cCoerce: ptr to: 'sqInt *') at: 0 put: val) to: 'sqInt'! Item was changed: ----- Method: MemoryAccess>>oopAt: (in category 'object memory access') ----- oopAt: oop "Answer the object memory word, an sqInt, located at an object memory location." "oopAt(sqInt oop) { return oopAtPointer(pointerForOop(oop)); }" + <inline: true> - self inline: true. ^ self oopAtPointer: (self pointerForOop: oop)! Item was changed: ----- Method: MemoryAccess>>oopAt:put: (in category 'object memory access') ----- oopAt: oop put: val "Store an object memory word, an sqInt, at the given object memory location." "sqInt oopAtput(sqInt oop, sqInt val) { return oopAtPointerput(pointerForOop(oop), val); }" + <inline: true> - self inline: true. ^ self oopAtPointer: (self pointerForOop: oop) put: val! Item was changed: ----- Method: MemoryAccess>>oopAtPointer: (in category 'map address to value') ----- oopAtPointer: ptr "Answer the object memory word, an sqInt, located at a machine address." "sqInt oopAtPointer(char *ptr) { return (sqInt)(*((sqInt *)ptr)) }" + <inline: true> + <var: #ptr type: 'char *'> - self inline: true. - self var: #ptr type: 'char *'. ^ (self cCoerce: ptr to: 'sqInt *') at: 0! Item was changed: ----- Method: MemoryAccess>>oopAtPointer:put: (in category 'map address to value') ----- oopAtPointer: ptr put: val "Store an object memory word, an sqInt, at the given machine address." "sqInt oopAtPointerput(char *ptr, sqInt val) { return (sqInt)(*((sqInt *)ptr)= (sqInt)val) }" + <inline: true> + <var: #ptr type: 'char *'> - self inline: true. - self var: #ptr type: 'char *'. ^ (self cCoerce: ptr to: 'sqInt *') at: 0 put: val! Item was changed: ----- Method: MemoryAccess>>oopForPointer: (in category 'memory mapping') ----- oopForPointer: ptr "Answer the oop located at ptr, a machine address. This method maps machine addresses (C pointers) to object memory locations. sqInt oopForPointer(char *ptr) { return (sqInt)(ptr - sqMemoryBase); }" + <inline: true> + <returnTypeC: 'usqInt'> + <var: #ptr type: 'char *'> - self inline: true. - self returnTypeC: 'usqInt'. - self var: #ptr type: 'char *'. ^ self cCoerce: ptr - self sqMemoryBaseAddress to: 'usqInt'! Item was changed: ----- Method: MemoryAccess>>pointerForOop: (in category 'memory mapping') ----- pointerForOop: oop "Answer the machine address of the object to which oop refers. This method maps object memory locations to their underlying machine addresses (C pointers). char *pointerForOop(usqInt oop) { return sqMemoryBase + oop; }" + <inline: true> + <returnTypeC: 'char *'> + <var: #oop type: 'usqInt'> - self inline: true. - self returnTypeC: 'char *'. - self var: #oop type: 'usqInt'. ^ self sqMemoryBaseAddress + oop! Item was changed: ----- Method: MemoryAccess>>shortAt: (in category 'object memory access') ----- shortAt: oop "Answer the signed short integer value at an object memory location. The result is a signed sqInt value. Negative values will be sign extended, such that if the short integer value is binary 16rFFFF, the result will be 16rFFFFFFFF for a 32-bit object memory, or 16rFFFFFFFFFFFFFFFF for a 64-bit object memory." "sqInt shortAt(sqInt oop) { return shortAtPointer(pointerForOop(oop)); }" + <inline: true> - self inline: true. ^ self shortAtPointer: (self pointerForOop: oop)! Item was changed: ----- Method: MemoryAccess>>shortAt:put: (in category 'object memory access') ----- shortAt: oop put: val "Cast a sqInt value to signed short integer and store its low order bytes at the given object memory location. Assuming that size of a short integer is 2, then two bytes are stored regardless of the size of sqInt (32-bit or 64-bit object memory)." "sqInt shortAtput(sqInt oop, int val){ return shortAtPointerput(pointerForOop(oop), val); }" + <inline: true> + <var: #val type: 'int'> - self inline: true. - self var: #val type: 'int'. ^ self shortAtPointer: (self pointerForOop: oop) put: val! Item was changed: ----- Method: MemoryAccess>>shortAtPointer: (in category 'map address to value') ----- shortAtPointer: ptr "Answer the signed short integer value at a machine address. The result is a signed sqInt value. Negative values will be sign extended, such that if the short integer value is binary 16rFFFF, the result will be 16rFFFFFFFF for a 32-bit object memory, or 16rFFFFFFFFFFFFFFFF for a 64-bit object memory." "sqInt shortAtPointer(char *ptr) { return (sqInt)(*((short *)ptr)) }" + <inline: true> + <var: #ptr type: 'char *'> - self inline: true. - self var: #ptr type: 'char *'. ^ self cCoerce: ((self cCoerce: ptr to: 'short *') at: 0) to: 'sqInt'! Item was changed: ----- Method: MemoryAccess>>shortAtPointer:put: (in category 'map address to value') ----- shortAtPointer: ptr put: val "Cast a sqInt value to signed short integer and store its low order bytes at the given machine address. Assuming that size of a short integer is 2, then two bytes are stored regardless of the size of sqInt (32-bit or 64-bit object memory)." "sqInt shortAtPointerput(char *ptr, int val) { return (sqInt)(*((short *)ptr)= (short)val) }" + <inline: true> + <var: #ptr type: 'char *'> + <var: #val type: 'int'> - self inline: true. - self var: #ptr type: 'char *'. - self var: # val type: 'int'. ^ self cCoerce: ((self cCoerce: ptr to: 'short *') at: 0 put: (self cCoerce: val to: 'short')) to: 'sqInt'! Item was changed: ----- Method: MemoryAccess>>shortUnsignedAtPointer: (in category 'map address to value') ----- shortUnsignedAtPointer: ptr "Answer the unsigned short integer value at a machine address. The result is a signed sqInt value with value in the range 0 through 16rFFFF, assuming that size of short integer is 2." "sqInt shortUnsignedAtPointer(char *ptr) { return (sqInt)(*((unsigned short *)ptr)) }" + <inline: true> + <var: #ptr type: 'char *'> - self inline: true. - self var: #ptr type: 'char *'. ^ self cCoerce: ((self cCoerce: ptr to: 'unsigned short *') at: 0) to: 'sqInt'! Item was changed: ----- Method: MemoryAccess>>sqMemoryBaseAddress (in category 'memory mapping') ----- sqMemoryBaseAddress "Answer the machine address of the start of object memory." + <inline: true> + <returnTypeC: 'char *'> - self inline: true. - self returnTypeC: 'char *'. ^ self cCode: 'sqMemoryBase' !

1 0

VM Maker: VMMaker.oscog-eem.1751.mcz
by commits＠source.squeak.org 31 Mar '16

31 Mar '16

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.1751.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.1751 Author: eem Time: 1 January 1970, 3:02:52.225589 pm UUID: 1ae8a7b7-5497-4eba-9511-198cc5aa7485 Ancestors: VMMaker.oscog-cb.1750 Revise the IMMUTABILITY store check to remove the need for the branch around the setting of TempReg with the inst var index. By having ceStoreTrampoline repeat the immutability test it doesn't care about the value of TempReg in the store check case and so we don't need to jump around the instruction that sets TempReg. This reduces the store check by a further 2 byte instruction. Clément, if you hate this we can back out of it, but I thought I'd commit it all the same. Simulator: Fix casting in the JIT simulator for the recent integer conversion routine modifications. =============== Diff against VMMaker.oscog-cb.1750 =============== Item was changed: ----- Method: CogObjectRepresentationForSpur>>genStoreTrampolineCalled: (in category 'initialization') ----- genStoreTrampolineCalled: trampolineName "This can be entered in one of two states, depending on TempReg. TempReg = 0 => store check TempReg > 0 => immutability failure TempReg holds index + 1 in this case as the value 0 is reserved for store checks. In addition the 0 value is convenient to save one instruction for store checks." | jumpSC | <option: #IMMUTABILITY> <var: #trampolineName type: #'char *'> <var: #jumpSC type: #'AbstractInstruction *'> <inline: false> cogit zeroOpcodeIndex. + cogit PushR: SendNumArgsReg. + jumpSC := self genJumpMutable: ReceiverResultReg scratchReg: SendNumArgsReg. + cogit PopR: SendNumArgsReg. - cogit CmpCq: 0 R: TempReg. - jumpSC := cogit JumpZero: 0. - - "CannotAssignTo:, we restore the index." - cogit SubCq: 1 R: TempReg. cogit compileTrampolineFor: #ceCannotAssignTo:withIndex:valueToAssign: numArgs: 3 arg: ReceiverResultReg arg: TempReg arg: ClassReg arg: nil regsToSave: cogit emptyRegisterMask pushLinkReg: true resultReg: NoReg. "Store check" + jumpSC jmpTarget: (cogit PopR: SendNumArgsReg). - jumpSC jmpTarget: cogit Label. ^ cogit genTrampolineFor: #remember: called: trampolineName numArgs: 1 arg: ReceiverResultReg arg: nil arg: nil arg: nil regsToSave: cogit emptyRegisterMask pushLinkReg: true resultReg: NoReg appendOpcodes: true! Item was changed: ----- Method: CogObjectRepresentationForSpur>>genStoreWithImmutabilityAndStoreCheckSourceReg:slotIndex:destReg:scratchReg:needRestoreRcvr: (in category 'compile abstract instructions') ----- genStoreWithImmutabilityAndStoreCheckSourceReg: sourceReg slotIndex: index destReg: destReg scratchReg: scratchReg needRestoreRcvr: needRestoreRcvr "Store check code is duplicated to use a single trampoline" + | immutableJump jmpImmediate jmpDestYoung jmpSourceOld rememberedBitByteOffset jmpAlreadyRemembered mask | <var: #immutableJump type: #'AbstractInstruction *'> - <var: #trampJump type: #'AbstractInstruction *'> <var: #jmpImmediate type: #'AbstractInstruction *'> <var: #jmpDestYoung type: #'AbstractInstruction *'> <var: #jmpSourceOld type: #'AbstractInstruction *'> <var: #jmpAlreadyRemembered type: #'AbstractInstruction *'> + - | immutableJump trampJump jmpImmediate jmpDestYoung jmpSourceOld rememberedBitByteOffset jmpAlreadyRemembered mask | - immutableJump := self genJumpImmutable: destReg scratchReg: scratchReg. cogit genTraceStores. "do the store" cogit MoveR: sourceReg Mw: index * objectMemory wordSize + objectMemory baseHeaderSize r: destReg. "store check" jmpImmediate := self genJumpImmediate: sourceReg. "Get the old/new boundary in scratchReg" cogit MoveCw: objectMemory storeCheckBoundary R: scratchReg. "Is target young? If so we're done" cogit CmpR: scratchReg R: destReg. "N.B. FLAGS := destReg - scratchReg" jmpDestYoung := cogit JumpBelow: 0. "Is value stored old? If so we're done." cogit CmpR: scratchReg R: sourceReg. "N.B. FLAGS := valueReg - scratchReg" jmpSourceOld := cogit JumpAboveOrEqual: 0. "value is young and target is old. Need to remember this only if the remembered bit is not already set. Test the remembered bit. Only need to fetch the byte containing it, which reduces the size of the mask constant." rememberedBitByteOffset := jmpSourceOld isBigEndian ifTrue: [objectMemory baseHeaderSize - 1 - (objectMemory rememberedBitShift // 8)] ifFalse:[objectMemory rememberedBitShift // 8]. mask := 1 << (objectMemory rememberedBitShift \\ 8). cogit MoveMb: rememberedBitByteOffset r: destReg R: scratchReg. cogit AndCq: mask R: scratchReg. jmpAlreadyRemembered := cogit JumpNonZero: 0. + "Set the inst var index for the benefit of the immutability check. The trampoline will + repeat the check to choose between the immutbality violation and the store check." + immutableJump jmpTarget: (cogit MoveCq: index R: scratchReg). + cogit CallRT: ceStoreTrampoline. - "We know scratchReg now holds 0, this is convenient because the trampoline - convention expects 0 for store check in scratchReg. What a coincidence ;-)" - "Remembered bit is not set. Call store check to insert dest into remembered table." - trampJump := cogit Jump: 0. - "Here we reach the trampoline for Immutability failure" - immutableJump jmpTarget: (cogit MoveCq: index + 1 R: scratchReg). "index + 1 as 0 is reserved for store checks" - trampJump jmpTarget: (cogit CallRT: ceStoreTrampoline). cogit annotateBytecode: cogit Label. needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. jmpImmediate jmpTarget: (jmpDestYoung jmpTarget: (jmpSourceOld jmpTarget: (jmpAlreadyRemembered jmpTarget: cogit Label))). ^ 0! Item was changed: ----- Method: CogObjectRepresentationForSpur>>genStoreWithImmutabilityButNoStoreCheckSourceReg:slotIndex:destReg:scratchReg:needRestoreRcvr: (in category 'compile abstract instructions') ----- genStoreWithImmutabilityButNoStoreCheckSourceReg: sourceReg slotIndex: index destReg: destReg scratchReg: scratchReg needRestoreRcvr: needRestoreRcvr <var: #mutableJump type: #'AbstractInstruction *'> <var: #immutabilityFailure type: #'AbstractInstruction *'> | immutabilityFailure mutableJump | "imm check has its own trampoline" mutableJump := self genJumpMutable: destReg scratchReg: scratchReg. + cogit MoveCq: index R: TempReg. - cogit MoveCq: index + 1 R: TempReg. "index + 1 as 0 is reserved for store checks" cogit CallRT: ceStoreTrampoline. cogit annotateBytecode: cogit Label. needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. immutabilityFailure := cogit Jump: 0. mutableJump jmpTarget: cogit Label. cogit genTraceStores. "do the store" cogit MoveR: sourceReg Mw: index * objectMemory wordSize + objectMemory baseHeaderSize r: destReg. immutabilityFailure jmpTarget: cogit Label. ^ 0! Item was changed: ----- Method: CogVMSimulator>>cCoerceSimple:to: (in category 'translation support') ----- cCoerceSimple: value to: cTypeString "Type coercion for translation and simulation. For simulation answer a suitable surrogate for the struct types" + ^cTypeString + caseOf: + { [#'CogMethod *'] -> [cogit cogMethodSurrogateAt: value asUnsignedInteger]. + [#'CogBlockMethod *'] -> [cogit cogBlockMethodSurrogateAt: value asUnsignedInteger]. + [#'NSSendCache *'] -> [cogit nsSendCacheSurrogateAt: value asUnsignedInteger] } + otherwise: [value]! - ^cTypeString caseOf: - { [#'char *'] -> [value]. - [#'void *'] -> [value]. - [#sqInt] -> [value]. - [#'void (*)()'] -> [value]. - [#'void (*)(void)'] -> [value]. - [#'CogMethod *'] -> [cogit cogMethodSurrogateAt: value asUnsignedInteger]. - [#'CogBlockMethod *'] -> [cogit cogBlockMethodSurrogateAt: value asUnsignedInteger]. - [#'NSSendCache *'] -> [cogit nsSendCacheSurrogateAt: value asUnsignedInteger]. }!

1 0

Attempting to move Cog to Git(Hub)
by Fabio Niephaus 30 Mar '16

30 Mar '16

Hi all, I've been playing around with Cog to see what it'd take to move the code to git. GitHub's svn importer [1] isn't very reliable I think (some directories were actually missing), but eventually I managed to convert the entire svn repo to git [2]. Also, I have been able to build different Cog VMs automatically on Travis CI. For a working build, see [3]. You can download the results at [4]. Please keep in mind that this is still a "test" ;) Eliot was also concerned how the version info will be maintained and I have a simple proposal: Each Cog release will be a git tag and for the version itself we could use semantic versioning [3]. Then the build version is the git tag, the date is the commit date and the URL is the "remote.origin.url". Here's an example: ==================================================================== 5.0-v0.1.0 Wed Mar 30 12:01:50 UTC 2016 gcc 4.6.3 [Production Spur 64-bit VM] CoInterpreter VMMaker.oscog-eem.1740 uuid: 10ece78d-d04d-4ad2-8598-e496b6a6097c Mar 30 2016 StackToRegisterMappingCogit VMMaker.oscog-eem.1740 uuid: 10ece78d-d04d-4ad2-8598-e496b6a6097c Mar 30 2016 VM: v0.1.0 https://github.com/fniephaus/squeak.git Date: Wed Mar 30 13:58:46 2016 +0200 Plugins: v0.1.0 https://github.com/fniephaus/squeak.git Linux testing-worker-linux-docker-9d959019-3394-linux-16 3.13.0-40-generic #69-Ubuntu SMP Thu Nov 13 17:53:56 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux plugin path: ./products/nscogspur64linuxht/lib/nsvm/5.0-v0.1.0 [default: /home/idea4it/products/nscogspur64linuxht/lib/nsvm/5.0-v0.1.0/] ==================================================================== Any further suggestions? :) Best, Fabio [1] https://import.github.com/ [2] https://github.com/fniephaus/squeak/tree/Cog [3] https://travis-ci.org/fniephaus/squeak/builds/119507180 [4] https://www.hpi.uni-potsdam.de/hirschfeld/artefacts/cog/v0.1.0/ [5] http://semver.org/

2 3

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev March 2016