Re: [Vm-dev] Serious inlining problem

Hi Nicolas,

I'm looking at this now. Your recent changes have also caused a huge number of warnings like this:

type mismatch for formal integerValue and actual "pwr - 1" when inlining pushInteger: in primitiveExponent. Use a cast. type mismatch for formal reasonCode and actual "(objectMemory isIntegerObject: index) ifTrue: [PrimErrBadIndex] ifFalse: [PrimErrBadArgument]" when inlining primitiveFailFor: in primitiveFloatAt. Use a cast. type mismatch for formal cond and actual "rcvr = arg" when inlining booleanCheat: in bytecodePrimIdentical. Use a cast. type mismatch for formal cond and actual "rcvr = arg" when inlining booleanCheatSistaV1: in bytecodePrimIdenticalSistaV1. Use a cast. type mismatch for formal cond and actual "rcvr = arg" when inlining booleanCheatV4: in bytecodePrimIdenticalV4. Use a cast. type mismatch for formal reasonCode and actual "(objectMemory isIndexable: rcvr) ifTrue: [PrimErrBadIndex] ifFalse: [PrimErrBadReceiver]" when inlining primitiveFailFor: in commonVariable:at:put:cacheIndex:. Use a cast. type mismatch for formal numSlots and actual "MessageLookupClassIndex + 1" when inlining eeInstantiateSmallClassIndex:format:numSlots: in createActualMessageTo:. Use a cast. type mismatch for formal successBoolean and actual "integerArg ~= 0" when inlining success: in doPrimitiveDiv:by:. Use a cast. type mismatch for formal successBoolean and actual "(result >> 60 + 1 bitAnd: 15) <= 1" when inlining success: in doPrimitiveDiv:by:. Use a cast. type mismatch for formal successBoolean and actual "integerArg ~= 0" when inlining success: in doPrimitiveMod:by:. Use a cast. type mismatch for formal successBoolean and actual "(integerResult >> 60 + 1 bitAnd: 15) <= 1" when inlining success: in doPrimitiveMod:by:. Use a cast. type mismatch for formal successBoolean and actual "arg ~= 0.0" when inlining success: in primitiveFloatDivide:byArg:. Use a cast. type mismatch for formal successBoolean and actual "objectsWritten = 1" when inlining success: in putLong:toFile:. Use a cast. type mismatch for formal successBoolean and actual "objectsWritten = 1" when inlining success: in putShort:toFile:. Use a cast. type mismatch for formal successBoolean and actual "objectsWritten = 1" when inlining success: in putWord32:toFile:. Use a cast.

:-(

We have to be careful. Pharo is trying to release. I would like a stable VM. I would also like a better-inlined VM, so I support what you're doing. I'm just worried about timing.

On Wed, Apr 27, 2016 at 12:47 PM, Nicolas Cellier < nicolas.cellier.aka.nice@gmail.com> wrote:

Hi, I changed the compiler flag -Wno-unused-value to -Wunused-value because there's no reason to carry dead code. Dead code should ring a bell. it seems it uncovers a problem in code generation. An example is in code generated for sendInvokeCallbackContext:

   if ((argumentCountOfMethodHeader(methodHeaderOf(GIV(newMethod))))

== 4) { /* begin push: */ value = ((usqInt)((vmCallbackContext->thunkp))); /* begin positive32BitIntegerFor: */ /* begin maybeInlinePositive32BitIntegerFor: */ assert(!((hasSixtyFourBitImmediates()))); if ((((unsigned int) value)) <= (MaxSmallInteger)) { ((value << 1) | 1); ^~~~~~~~~~~~~~ PROBLEM HERE: this result should be stored in variable 'object' goto l4; } /* begin eeInstantiateSmallClassIndex:format:numSlots: */ ...snip... object = newLargeInteger; l4: /* end maybeInlinePositive32BitIntegerFor: */; goto l5;

    l5:     /* end positive32BitIntegerFor: */;

            longAtput((sp = GIV(stackPointer) - BytesPerWord), object);

The corresponding Slang code is:

(self argumentCountOf: newMethod) = 4 ifTrue:
    [self push: (self positiveMachineIntegerFor: vmCallbackContext

thunkp asUnsignedInteger).

With push:

| sp |
<inline: true>
<var: #sp type: #'char *'>
stackPages longAt: (sp := stackPointer - objectMemory wordSize) put:

object. stackPointer := sp

With positiveMachineIntegerFor:

<var: #value type: #'unsigned long'>
<inline: true>
^objectMemory wordSize = 8
    ifTrue: [self positive64BitIntegerFor: value]
    ifFalse: [self positive32BitIntegerFor: value]

With positive32BitIntegerFor:

<inline: true>
<var: 'integerValue' type: #'unsigned int'>
objectMemory hasSixtyFourBitImmediates
    ifTrue:
        [^objectMemory integerObjectOf: (integerValue asUnsignedLong

bitAnd: 16rFFFFFFFF)] ifFalse: [^self maybeInlinePositive32BitIntegerFor: integerValue]

With maybeInlinePositive32BitIntegerFor:

<notOption: #Spur64BitMemoryManager>
<var: 'integerValue' type: #'unsigned int'>
| newLargeInteger |
self deny: objectMemory hasSixtyFourBitImmediates.
   "force coercion because slang inliner sometimes incorrectly pass a

signed int without converting to unsigned" (self cCode: [self cCoerceSimple: integerValue to: #'unsigned int'] inSmalltalk: [integerValue bitAnd: 1 << 32 - 1]) <= objectMemory maxSmallInteger ifTrue: [^objectMemory integerObjectOf: integerValue]. newLargeInteger := objectMemory eeInstantiateSmallClassIndex: ...snip... ^newLargeInteger