Removing duplicated code

[jordanmontt]

Moved #moveFramesIn:through:toPage: method to the super class and added hook methods to specialize in the subclasses (#defaultCallerSavedIP, #getCallerIPFromFP:callerFP:).
Moved #genPushRegisterArgsForAbortMissNumArgs: to the new super class CogX86Compiler.
Moved #changeClassOf:to: method to the super class and rewrote it in terms of the word size.
Added a new test class VMPrimitiveChangeClassParametrizedTest

[jordanmontt]

Done in the Vm Dojo with @PalumboN @guillep @fouziray @iglosiggio @FedeLoch

[iglosiggio]

I think we can do better:

Suggested change

	{ #category : 'interpreter access' }
	SpurMemoryManager >> paddingForFormatInstBytes: instBytes forElementBytes: elementSizeInBytes shiftForElement: shiftForElement [
	"This method is doing the same as below but using bit arithmetic.
	shiftForElement SHOULD be log 2 of element size in bytes. We don't do the log2 because 1) is slow 2) it can return floats
	slotsPerWord := (wordSize / elementSize).
	trailingBytes := instBytes \\ wordSize.
	trailingSlots := trailingBytes / elementSize.
	^ slotsPerWord - trailingSlots. "
	^ self wordSize // elementSizeInBytes - (instBytes >> shiftForElement) bitAnd: self wordSize // elementSizeInBytes - 1
	]
	{ #category : 'interpreter access' }
	SpurMemoryManager >> paddingForFormatInstBytes: instBytes shiftForElement: shiftForElement [
	"NOTE: shiftForElement is elementSizeInBytes log: 2.
	This means elementSizeInBytes should be a power of two."
	| trailingBytes paddingBytes |
	"Cheap way to do instBytes \\ self wordSize (assumes self wordSize is a power of two)"
	trailingBytes := instBytes bitAnd: self wordSize - 1.
	paddingBytes := self wordSize - trailingBytes.
	"self wordSize - 0 is self wordSize, but we want to say 'there is no padding' in that case"
	paddingBytes := paddingBytes bitAnd: self wordSize - 1.
	"Cheap way to do paddingBytes // elementSizeInBytes"
	^ paddingBytes >> shiftForElement.
	]

[iglosiggio]

NOTE: this change removes the elementBytes parameter :P

[iglosiggio]

There's a bug here when self wordSize < elementSizeInBytes (in those cases we never have padding).

The ugly fix is to add an if returning zero on that case.