Vm-dev August 2015

vm-dev@lists.squeakfoundation.org

37 participants
78 discussions

Re: [Pharo-dev] [Moose-dev] Re: Object new foo
by Eliot Miranda 25 Aug '15

25 Aug '15

Hi Vincent, On Sun, Aug 23, 2015 at 9:53 AM, Vincent BLONDEAU < vincent.blondeau(a)polytech-lille.net> wrote: > Hi Eliot, > > > > Thanks for your help. > > The problem has been resolved! > > In fact a class was removed or renamed in Pharo, and the Moose debugger > calls it just before the opening. > > So it created an infinite loop, because the class wasn’t understood by the > system, so it opened another debugger, etc…. > > > > Do you think something can be made to not break the image and so to break > this infinite loop? > It may make sense to set a flag early in Debugger opening, and test this flag immediately before it is set to check for recursion. But the issue is what to do in that case. One can't open a debugger, because the attempt is failing. So one can only fall back on some more primitive error handler. Note that the flag should be specific to the thread of control, which may be difficult if the Debugger runs in a separate process. One doesn't want to prevent multiple debuggers, only stop recursion within a single error. > A solution may to open either another registered debugger, or the > emergency debugger? > Yes, the emergency debugger. > > > I can create an image with the problem if needed. > > > > Vincent > > > > *De :* Pharo-dev [mailto:pharo-dev-bounces@lists.pharo.org] *De la part > de* Eliot Miranda > *Envoyé :* jeudi 20 août 2015 22:13 > *À :* Pharo Development List > *Cc :* Moose-related development; Squeak Virtual Machine Development > Discussion > *Objet :* Re: [Pharo-dev] [Moose-dev] Re: Object new foo > > > > Hi Vincent, > > > > does the image lock up when you execute Exception signal? The two > functions you see looping are part of the implementation of primitives 195 > & 196, findNextUnwindContextUpTo: (primitiveFindNextUnwindContext) > & terminateTo: (primitiveTerminateTo). So if the image locks up the > question is, is what is creating a circular structure of contexts, the VM > or the image? I am willing to take a look at an image that contains the > case. If you can create an image which enters the lop on start-up without > my having to do anything I'm happy to take a look. > > > > On Wed, Aug 19, 2015 at 3:22 PM, Vincent BLONDEAU < > vincent.blondeau(a)polytech-lille.net> wrote: > > Hi, > > > > On last Moose image, I debugged the VM and executed “Exception signal”. > > I paused the VM and saw that 2 methods are looping together in > gcc3x-cointerp.c: > > static sqInt findMethodWithPrimitiveFromContextUpToContext(sqInt > primitive, sqInt senderContext, sqInt homeContext); > > static sqInt findMethodWithPrimitiveFromFPUpToContext(sqInt primitive, > char *startFP, sqInt homeContext); > > > > The image is looping only when the debugger is launch on Error (not by > launching it by the script below). > > > > Maybe someone on the Pharo mailing list knows what is happening? > > > > Thanks in advance, > > > > Cheers, > > Vincent > > > > *De :* moose-dev-bounces(a)iam.unibe.ch [mailto: > moose-dev-bounces(a)iam.unibe.ch] *De la part de* Andrei Chis > *Envoyé :* mardi 18 août 2015 10:53 > *À :* Moose-related development > *Objet :* [Moose-dev] Re: Object new foo > > > > Hi Vincent, > > > > You example will also kill the Pharo debugger :) > > The issue is that you pass 'Processor activeProcess' which gets terminated > when you close the debugger. > > When you open the debugger you need to make sure the process > > on which it is opened is suspended and can get terminated with no problem. > > > > A different way to redo the example would be: > > > > context := Context > > sender: nil > > receiver: {1 . 2 . 2} > > method: (OrderedCollection>>#add:) > > arguments: #(10). > > process := Process > > forContext: context > > priority: Processor userInterruptPriority. > > GTGenericStackDebugger openOn: (DebugSession process: process context: > context). > > > > There is a basic test for opening the debugger > in GTGenericStackDebuggerSmokeTest but more could help. > > > > Cheers, > > Andrei > > > > > > On Mon, Aug 17, 2015 at 11:53 PM, Vincent BLONDEAU < > vincent.blondeau(a)polytech-lille.net> wrote: > > Indeed. Maybe some tests has to be added.. > > > > But for now, it seems that the debugger is crashing the image… Not just by > launching it: > > GTGenericStackDebugger openOn: (DebugSession process: Processor > activeProcess context: (Context > > sender: nil > > receiver: {1 . 2 . 2} > > method: (OrderedCollection>>#add: ) > > arguments: #(10))) > > > > But when an exception is thrown, its seems that we fall into an infinite > loop around the primitive 199. Maybe the context is not well set? > > > > Cheers, > > Vincent > > > > > > > > *De :* moose-dev-bounces(a)iam.unibe.ch [mailto: > moose-dev-bounces(a)iam.unibe.ch] *De la part de* Alexandre Bergel > *Envoyé :* lundi 17 août 2015 13:26 > *À :* Moose-related development > *Objet :* [Moose-dev] Re: Object new foo > > > > This is weird that such bug can be introduced and no alarm is raised > > > > Alexandre > > > Le 17 août 2015 à 08:03, Jan Kurš <kurs(a)iam.unibe.ch> a écrit : > > Yep I had this problem (falled back to previous version), I suspected > debugger being the reason. > > Cheers Jan > > > > On Sun, 16 Aug 2015 03:05 Alexandre Bergel <alexandre.bergel(a)me.com> > wrote: > > in Moose 6.0 simply freezes. > Anyone else sees this? > > Alexandre > -- > _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: > Alexandre Bergel http://www.bergel.eu > ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > > > > _______________________________________________ > Moose-dev mailing list > Moose-dev(a)iam.unibe.ch > https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > _______________________________________________ > Moose-dev mailing list > Moose-dev(a)iam.unibe.ch > https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > > _______________________________________________ > Moose-dev mailing list > Moose-dev(a)iam.unibe.ch > https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > > > > > > > -- > > _,,,^..^,,,_ > > best, Eliot > -- _,,,^..^,,,_ best, Eliot

2 1

VM Maker: VMMaker.oscog-rmacnak.1443.mcz
by commits＠source.squeak.org 25 Aug '15

25 Aug '15

Ryan Macnak uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-rmacnak.1443.mcz ==================== Summary ==================== Name: VMMaker.oscog-rmacnak.1443 Author: rmacnak Time: 24 August 2015, 11:09:53.273 pm UUID: fd1228e3-f910-43a8-b648-f197510852c5 Ancestors: VMMaker.oscog-rmacnak.1442 Further fix outer send MNU by not confusing lkupClass and lkupClassTag. Push setting lkupClassTag down to clarify it is for new cache entries. =============== Diff against VMMaker.oscog-rmacnak.1442 =============== Item was changed: ----- Method: CoInterpreter>>ceDynamicSuperSend:receiver: (in category 'trampolines') ----- ceDynamicSuperSend: cacheAddress receiver: methodReceiver "A dynamic super send cache missed." | nsSendCache methodReceiverClassTag cogMethod errSelIdx | <api> <option: #NewspeakVM> <inline: false> <var: #nsSendCache type: #'NSSendCache *'> <var: #cogMethod type: #'CogMethod *'> cogit assertCStackWellAligned. self assert: (objectMemory addressCouldBeOop: methodReceiver). self deny: (objectMemory isOopForwarded: methodReceiver). nsSendCache := self cCoerceSimple: cacheAddress to: #'NSSendCache *'. messageSelector := nsSendCache selector. argumentCount := nsSendCache numArgs. method := (self mframeHomeMethod: framePointer) methodObject. self assert: (self stackValue: argumentCount + 1 "ret addr") = methodReceiver. methodReceiverClassTag := objectMemory fetchClassTagOf: methodReceiver. (self inlineLookupInNSMethodCacheSel: messageSelector classTag: methodReceiverClassTag method: method lookupRule: LookupRuleDynamicSuper) ifTrue: [self assert: localAbsentReceiverOrZero = 0. "check for coggability because method is in the cache" self ifAppropriateCompileToNativeCode: newMethod selector: messageSelector] ifFalse: [self deny: (objectMemory isOopForwarded: messageSelector). self deny: (objectMemory isForwardedClassTag: methodReceiverClassTag). - lkupClassTag := methodReceiverClassTag. errSelIdx := self lookupDynamicSuperSendNoMNU: methodReceiver. errSelIdx ~= 0 ifTrue: [self assert: errSelIdx == SelectorDoesNotUnderstand. + self assert: localAbsentReceiver == methodReceiver. self handleMNU: errSelIdx InMachineCodeTo: methodReceiver + classForMessage: lkupClass. - classForMessage: (objectMemory classForClassTag: methodReceiverClassTag). self unreachable]. + lkupClassTag := methodReceiverClassTag. self addNewMethodToNSCache: LookupRuleDynamicSuper]. (self maybeMethodHasCogMethod: newMethod) ifTrue: [ cogMethod := self cogMethodOf: newMethod. cogMethod selector = objectMemory nilObject ifTrue: [cogit setSelectorOf: cogMethod to: messageSelector] ifFalse: ["Deal with anonymous accessors, e.g. in Newspeak. The cogMethod may not have the correct selector. If not, try and compile a new method with the correct selector." cogMethod selector ~= messageSelector ifTrue: [ (cogit cog: newMethod selector: messageSelector) ifNotNil: [:newCogMethod | cogMethod := newCogMethod]]]. cogMethod selector = messageSelector ifTrue: [cogit linkNSSendCache: nsSendCache classTag: methodReceiverClassTag enclosingObject: 0 target: cogMethod caller: self mframeHomeMethodExport] ifFalse: ["Out of code memory. Fall through to interpret."]. instructionPointer := self popStack. self executeNewMethod. self unreachable]. instructionPointer := self popStack. self interpretMethodFromMachineCode. self unreachable.! Item was changed: ----- Method: CoInterpreter>>ceImplicitReceiverSend:receiver: (in category 'trampolines') ----- ceImplicitReceiverSend: cacheAddress receiver: methodReceiver "An implicit receiver send cache missed." | nsSendCache methodReceiverClassTag cogMethod errSelIdx | <api> <option: #NewspeakVM> <inline: false> <var: #nsSendCache type: #'NSSendCache *'> <var: #cogMethod type: #'CogMethod *'> cogit assertCStackWellAligned. self assert: (objectMemory addressCouldBeOop: methodReceiver). self deny: (objectMemory isOopForwarded: methodReceiver). nsSendCache := self cCoerceSimple: cacheAddress to: #'NSSendCache *'. messageSelector := nsSendCache selector. argumentCount := nsSendCache numArgs. method := (self mframeHomeMethod: framePointer) methodObject. self assert: (self stackValue: argumentCount + 1 "ret addr") = methodReceiver. methodReceiverClassTag := objectMemory fetchClassTagOf: methodReceiver. (self inlineLookupInNSMethodCacheSel: messageSelector classTag: methodReceiverClassTag method: method lookupRule: LookupRuleImplicit) ifTrue: [localAbsentReceiverOrZero = 0 ifTrue: [localAbsentReceiver := methodReceiver] ifFalse: [localAbsentReceiver := localAbsentReceiverOrZero]. "check for coggability because method is in the cache" self ifAppropriateCompileToNativeCode: newMethod selector: messageSelector] ifFalse: [self deny: (objectMemory isOopForwarded: messageSelector). self deny: (objectMemory isForwardedClassTag: methodReceiverClassTag). - lkupClassTag := methodReceiverClassTag. errSelIdx := self lookupImplicitReceiverSendNoMNU: methodReceiver. errSelIdx ~= 0 ifTrue: [self assert: errSelIdx == SelectorDoesNotUnderstand. + self assert: localAbsentReceiver == methodReceiver. self handleMNU: errSelIdx InMachineCodeTo: methodReceiver + classForMessage: lkupClass. - classForMessage: (objectMemory classForClassTag: methodReceiverClassTag). self unreachable]. + lkupClassTag := methodReceiverClassTag. self addNewMethodToNSCache: LookupRuleImplicit]. "Fix stacked receiver." self stackValue: argumentCount + 1 "ret addr" put: localAbsentReceiver. (self maybeMethodHasCogMethod: newMethod) ifTrue: [ cogMethod := self cogMethodOf: newMethod. cogMethod selector = objectMemory nilObject ifTrue: [cogit setSelectorOf: cogMethod to: messageSelector] ifFalse: ["Deal with anonymous accessors, e.g. in Newspeak. The cogMethod may not have the correct selector. If not, try and compile a new method with the correct selector." cogMethod selector ~= messageSelector ifTrue: [ (cogit cog: newMethod selector: messageSelector) ifNotNil: [:newCogMethod | cogMethod := newCogMethod]]]. cogMethod selector = messageSelector ifTrue: [cogit linkNSSendCache: nsSendCache classTag: methodReceiverClassTag enclosingObject: localAbsentReceiverOrZero target: cogMethod caller: self mframeHomeMethodExport] ifFalse: ["Out of code memory. Fall through to interpret."]. instructionPointer := self popStack. self executeNewMethod. self unreachable]. instructionPointer := self popStack. self interpretMethodFromMachineCode. self unreachable.! Item was changed: ----- Method: CoInterpreter>>ceOuterSend:receiver: (in category 'trampolines') ----- ceOuterSend: cacheAddress receiver: methodReceiver "An outer send cache missed." | nsSendCache depth methodReceiverClassTag cogMethod errSelIdx | <api> <option: #NewspeakVM> <inline: false> <var: #nsSendCache type: #'NSSendCache *'> <var: #cogMethod type: #'CogMethod *'> cogit assertCStackWellAligned. self assert: (objectMemory addressCouldBeOop: methodReceiver). self deny: (objectMemory isOopForwarded: methodReceiver). nsSendCache := self cCoerceSimple: cacheAddress to: #'NSSendCache *'. messageSelector := nsSendCache selector. argumentCount := nsSendCache numArgs. depth := nsSendCache depth. method := (self mframeHomeMethod: framePointer) methodObject. self assert: (self stackValue: argumentCount + 1 "ret addr") = methodReceiver. methodReceiverClassTag := objectMemory fetchClassTagOf: methodReceiver. (self inlineLookupInNSMethodCacheSel: messageSelector classTag: methodReceiverClassTag method: method lookupRule: depth) ifTrue: [localAbsentReceiverOrZero = 0 ifTrue: [localAbsentReceiver := methodReceiver] ifFalse: [localAbsentReceiver := localAbsentReceiverOrZero]. "check for coggability because method is in the cache" self ifAppropriateCompileToNativeCode: newMethod selector: messageSelector] ifFalse: [self deny: (objectMemory isOopForwarded: messageSelector). self deny: (objectMemory isForwardedClassTag: methodReceiverClassTag). - lkupClassTag := methodReceiverClassTag. "Successful lookup cached under the sender's class" errSelIdx := self lookupOuterSendNoMNU: methodReceiver depth: depth. errSelIdx ~= 0 ifTrue: [self assert: errSelIdx == SelectorDoesNotUnderstand. "Fix stacked receiver." self stackValue: argumentCount + 1 "ret addr" put: localAbsentReceiver. - "Failed lookup continues with the enclosing object's class" - lkupClassTag := objectMemory classForClassTag: localAbsentReceiver. self handleMNU: errSelIdx InMachineCodeTo: localAbsentReceiver + classForMessage: lkupClass. - classForMessage: lkupClassTag. self unreachable]. + lkupClassTag := methodReceiverClassTag. "Successful lookup cached under the sender's class" self addNewMethodToNSCache: depth]. "Fix stacked receiver." self stackValue: argumentCount + 1 "ret addr" put: localAbsentReceiver. (self maybeMethodHasCogMethod: newMethod) ifTrue: [ cogMethod := self cogMethodOf: newMethod. cogMethod selector = objectMemory nilObject ifTrue: [cogit setSelectorOf: cogMethod to: messageSelector] ifFalse: ["Deal with anonymous accessors, e.g. in Newspeak. The cogMethod may not have the correct selector. If not, try and compile a new method with the correct selector." cogMethod selector ~= messageSelector ifTrue: [ (cogit cog: newMethod selector: messageSelector) ifNotNil: [:newCogMethod | cogMethod := newCogMethod]]]. cogMethod selector = messageSelector ifTrue: [cogit linkNSSendCache: nsSendCache classTag: methodReceiverClassTag enclosingObject: localAbsentReceiverOrZero target: cogMethod caller: self mframeHomeMethodExport] ifFalse: ["Out of code memory. Fall through to interpret."]. instructionPointer := self popStack. self executeNewMethod. self unreachable]. instructionPointer := self popStack. self interpretMethodFromMachineCode. self unreachable.! Item was changed: ----- Method: CoInterpreter>>ceSelfSend:receiver: (in category 'trampolines') ----- ceSelfSend: cacheAddress receiver: methodReceiver "A self send cache missed." | nsSendCache methodReceiverClassTag cogMethod errSelIdx | <api> <option: #NewspeakVM> <inline: false> <var: #nsSendCache type: #'NSSendCache *'> <var: #cogMethod type: #'CogMethod *'> cogit assertCStackWellAligned. self assert: (objectMemory addressCouldBeOop: methodReceiver). self deny: (objectMemory isOopForwarded: methodReceiver). nsSendCache := self cCoerceSimple: cacheAddress to: #'NSSendCache *'. messageSelector := nsSendCache selector. argumentCount := nsSendCache numArgs. method := (self mframeHomeMethod: framePointer) methodObject. self assert: (self stackValue: argumentCount + 1 "ret addr") = methodReceiver. methodReceiverClassTag := objectMemory fetchClassTagOf: methodReceiver. (self inlineLookupInNSMethodCacheSel: messageSelector classTag: methodReceiverClassTag method: method lookupRule: LookupRuleSelf) ifTrue: [self assert: localAbsentReceiverOrZero = 0. "check for coggability because method is in the cache" self ifAppropriateCompileToNativeCode: newMethod selector: messageSelector] ifFalse: [self deny: (objectMemory isOopForwarded: messageSelector). self deny: (objectMemory isForwardedClassTag: methodReceiverClassTag). - lkupClassTag := methodReceiverClassTag. errSelIdx := self lookupOuterSendNoMNU: methodReceiver depth: 0. errSelIdx ~= 0 ifTrue: [self assert: errSelIdx == SelectorDoesNotUnderstand. + self assert: localAbsentReceiver == methodReceiver. self handleMNU: errSelIdx InMachineCodeTo: methodReceiver + classForMessage: lkupClass. - classForMessage: (objectMemory classForClassTag: methodReceiverClassTag). self unreachable]. + lkupClassTag := methodReceiverClassTag. self addNewMethodToNSCache: LookupRuleSelf]. (self maybeMethodHasCogMethod: newMethod) ifTrue: [ cogMethod := self cogMethodOf: newMethod. cogMethod selector = objectMemory nilObject ifTrue: [cogit setSelectorOf: cogMethod to: messageSelector] ifFalse: ["Deal with anonymous accessors, e.g. in Newspeak. The cogMethod may not have the correct selector. If not, try and compile a new method with the correct selector." cogMethod selector ~= messageSelector ifTrue: [ (cogit cog: newMethod selector: messageSelector) ifNotNil: [:newCogMethod | cogMethod := newCogMethod]]]. cogMethod selector = messageSelector ifTrue: [cogit linkNSSendCache: nsSendCache classTag: methodReceiverClassTag enclosingObject: 0 target: cogMethod caller: self mframeHomeMethodExport] ifFalse: ["Out of code memory. Fall through to interpret."]. instructionPointer := self popStack. self executeNewMethod. self unreachable]. instructionPointer := self popStack. self interpretMethodFromMachineCode. self unreachable.!

1 0

Slang & FFI
by Ben Coman 24 Aug '15

24 Aug '15

Just had a curious thought I could not find the answer from poking around... Is Slang able to make use of the FFI interface? I see things like file primitives with C-code scaffolding around system calls, and wonder if this C-code could be reduced by Slang calling directly to system functions through FFI, with scaffolding written in Slang? Does that make any sense? cheers -ben

3 2

Re: [Pharo-dev] issue with large file path on windows
by Ben Coman 24 Aug '15

24 Aug '15

(CC'd vm-dev) On Mon, Aug 24, 2015 at 12:09 AM, Nicolai Hess <nicolaihess(a)web.de> wrote: > And If you want to review the changes: > > https://github.com/nicolaihess/pharo-vm/compare/master...nicolaihess:win-lo… I don't have a Windows machine to test on, but I perused the changes. one spelling defect ../platforms/win32/vm/sqWin32.h, line 452 "functiosn" Now my curiosity is aroused regarding sqFileDeleteNameSize() in [1]. Where does its parameter "char* fileNameIndex" get its null termination? I can only guess maybe the following definition in primitiveFileDelete.st [2] <var: 'nameIndex' type: 'char *'> automatically appends a null to the Smalltalk string. My understanding is that the Smalltalk string's internal representation is not null terminated. I ask since the sqFileDeleteNameSize()'s comment "convert the file name into a null-terminated C string" indicates the null is added by MultiByteToWideChar(), but [3] indicates that is not true -- that the output of MultiByteToWideChar() is only null terminated if the input is null terminated. Also, anyone know if the concern "BIG WARNING about the return value and the `cbMultiByte` parameter" expressed near the end of [3] is relevant? [1] ../pharo-vm/platforms/win32/plugins/FilePlugin/sqWin32FilePrims.c [2] ../pharo-vm/mc/VMMaker.oscog.package/FilePlugin.class/instance/primitiveFileDelete.st [3] https://msdn.microsoft.com/en-us/library/windows/desktop/dd319072(v=vs.85).…

2 3

Re: [Pharo-users] issue with large file path on windows
by Ben Coman 24 Aug '15

24 Aug '15

On Mon, Aug 24, 2015 at 8:44 PM, Nicolai Hess <nicolaihess(a)web.de> wrote: > > > 2015-08-24 13:34 GMT+02:00 Henrik Johansen <henrik.s.johansen(a)veloxit.no>: > >> >> On 23 Aug 2015, at 6:09 , Nicolai Hess <nicolaihess(a)web.de> wrote: >> >> And If you want to review the changes: >> >> >> https://github.com/nicolaihess/pharo-vm/compare/master...nicolaihess:win-lo… >> >> >> 2015-08-23 13:44 GMT+02:00 Nicolai Hess <nicolaihess(a)web.de>: >> >>> For those who had problems with pharo on windows and github based >>> repositories, >>> I built a windows vm with support for long paths: >>> >>> >>> https://drive.google.com/file/d/0B8yEahnuIem2bmxwdzJuUXFxVGM/view?usp=shari… >>> >>> >>> For browsing directories with large paths (FileList or Inspect), >>> you may need one additional change in the image (But I am not really >>> sure about that) : >>> >>> DiskStore>>initialize >>> super initialize. >>> maxFileNameLength := Smalltalk vm maxFilenameLength ifNil: [ 32767 ]. >>> >>> >>> please test and give feedback. >>> >>> This wasn't as easy as I thought, and I had to make some more changes >>> for the file permissions (the stat-functions don't work for files with >>> long paths). >>> Please test other file/folder operations. >>> >>> >> - Is an alloca version really worth it for the potential problems? >> > > It is what Marcel used for the long path support in the squeak vm (I would > have copy and paste the whole change, but I couldn't find the appropriate > change for directory functions). > > "Potential problems" sounded like something interesting I could learn. Here is a summary... http://stackoverflow.com/questions/5807612/on-the-use-and-abuse-of-alloca On the use and abuse of alloca * alloca five times faster than malloc (but is that necessary given file operations likely take longer anyway?) * most concerns with alloca no problem here e.g. its a very short lifespan; the pointer is not returned from the function. http://www.strchr.com/alloca The Perils of alloca() * remaining concern is potential stack exhaustion... * is MAX_LONG_FILE_PATH=32kB significant today (but also consider embedded application) ? * counter argument, primitive functions are not likely to entered recursively ? Microsoft advises alloca [1] deprecated in favour of malloca [2] which allocates either stack or heap depending on requested size [1] https://msdn.microsoft.com/en-us/library/wb1s57t5(v=vs.80).aspx [2] https://msdn.microsoft.com/en-us/library/5471dc8s(VS.80).aspx https://sourceware.org/ml/libc-alpha/2012-06/msg00246.html An example of a DIY malloca... * might have the advantage of tuning MAX_ALLOCA_CUTOFF post-compilation * minGW gcc might not have malloca * if large allocations go on the heap, maybe MAX_LONG_FILE_PATH test not required, just MAX_ALLOCA_CUTOFF ---------- btw, perhaps rename CONVERT_MULTIBYTE_TO_WIDECHAR_PATH to ALLOC_WIDECHAR_PATH_FROM_MULTIBYTE or similar so its more obviously symmetrical with FREE_WIDECHAR_PATH. cheers -ben

1 0

Re: [Cuis] New Cog VMs available...
by Ken.Dickey 24 Aug '15

24 Aug '15

Hi, I am running Cuis on an ARM Chromebook using a superannuated cog stack-vm. Cuis has not yet migrated to Spur. Any chance of adding stklinuxARM to the build? Thanks a bunch, -KenD

1 0

New Cog VMs available...
by Eliot Miranda 24 Aug '15

24 Aug '15

... at http://www.mirandabanda.org/files/Cog/VM/VM.r3427. Squeak V5 users will want to upgrade their VMs because they, along with Smalltalk changes to follow soon, fix image segments. But upgrading is not a trivial process because the VMs on my site are not complete. The best way to update is to take a copy of the Squeak 5.0 all-in-one and replace the main VM executable there-in with one from my site. This gets you up-to-date plugins and an up-to-date VM. I hope that this process will get easier soon. ------------------------------------------------------------------------ CogVM binaries as per VMMaker.oscog-eem.1441/r3427 Modify Spur ImageSegment load to become the segmentWordArray into an Array of the loaded objects if load is successful, hence decoupling ImageSegment from the assumption that objects are allocated in order. Fix Integer receiver, float arg comparison with NaNs in the machine-code primitive. This has started failing in the FloatTest>>testNaNCompare since the new machine-code perform primitive invoked the machine-code version of the primitive. The Interpretewr code has always been correct and the old perform primitive would always run the Interpreter primitive if it exsted, since this would probably be faster. Fix the bug introduced by the fix to primitive function invocation in VMMaker.oscog-eem.1351 The fix correctly changed primitve code to set the primitiveFunctionPointer appropriately when a jitted external primitive was rebound, but it didn't remember to void the jit's record of the offset of the assignment that sets the primitiveFunctionPointer when switching between profiling andf non-profiling regimes, so that the address from the wrong regime would remain and be used to smash prmitive code. The fix is simply to void the externalSetPrimOffsets in voidCogCompiledCode. This fixes the bug whose symptom is a hard VM crash when using AndreasSystemProfilier. Integrate Marcel Taeumel & Tobias Pape's v2 SSL plugin changes. Fix negative 64-bit shift in the 64-bit Spur Stack interpreter. Newspeak: Fix MNU for cogged self and outer sends. Make the Newspeak VM packager include the V50 sources file instead of V41. _,,,^..^,,,_ best, Eliot

4 5

VM Maker: VMMaker.oscog-rmacnak.1442.mcz
by commits＠source.squeak.org 23 Aug '15

23 Aug '15

Ryan Macnak uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-rmacnak.1442.mcz ==================== Summary ==================== Name: VMMaker.oscog-rmacnak.1442 Author: rmacnak Time: 22 August 2015, 4:28:54.585 pm UUID: f34e246e-855a-4049-bcc7-76440e7a5e54 Ancestors: VMMaker.oscog-eem.1441 The actual receiver of an MNU from an outer send is not necessarily the method receiver. =============== Diff against VMMaker.oscog-eem.1441 =============== Item was changed: ----- Method: CoInterpreter>>ceOuterSend:receiver: (in category 'trampolines') ----- ceOuterSend: cacheAddress receiver: methodReceiver "An outer send cache missed." | nsSendCache depth methodReceiverClassTag cogMethod errSelIdx | <api> <option: #NewspeakVM> <inline: false> <var: #nsSendCache type: #'NSSendCache *'> <var: #cogMethod type: #'CogMethod *'> cogit assertCStackWellAligned. self assert: (objectMemory addressCouldBeOop: methodReceiver). self deny: (objectMemory isOopForwarded: methodReceiver). nsSendCache := self cCoerceSimple: cacheAddress to: #'NSSendCache *'. messageSelector := nsSendCache selector. argumentCount := nsSendCache numArgs. depth := nsSendCache depth. method := (self mframeHomeMethod: framePointer) methodObject. self assert: (self stackValue: argumentCount + 1 "ret addr") = methodReceiver. methodReceiverClassTag := objectMemory fetchClassTagOf: methodReceiver. (self inlineLookupInNSMethodCacheSel: messageSelector classTag: methodReceiverClassTag method: method lookupRule: depth) ifTrue: [localAbsentReceiverOrZero = 0 ifTrue: [localAbsentReceiver := methodReceiver] ifFalse: [localAbsentReceiver := localAbsentReceiverOrZero]. "check for coggability because method is in the cache" self ifAppropriateCompileToNativeCode: newMethod selector: messageSelector] ifFalse: [self deny: (objectMemory isOopForwarded: messageSelector). self deny: (objectMemory isForwardedClassTag: methodReceiverClassTag). + lkupClassTag := methodReceiverClassTag. "Successful lookup cached under the sender's class" - lkupClassTag := methodReceiverClassTag. errSelIdx := self lookupOuterSendNoMNU: methodReceiver depth: depth. errSelIdx ~= 0 ifTrue: [self assert: errSelIdx == SelectorDoesNotUnderstand. + "Fix stacked receiver." + self stackValue: argumentCount + 1 "ret addr" put: localAbsentReceiver. + "Failed lookup continues with the enclosing object's class" + lkupClassTag := objectMemory classForClassTag: localAbsentReceiver. self handleMNU: errSelIdx + InMachineCodeTo: localAbsentReceiver + classForMessage: lkupClassTag. - InMachineCodeTo: methodReceiver - classForMessage: (objectMemory classForClassTag: methodReceiverClassTag). self unreachable]. self addNewMethodToNSCache: depth]. "Fix stacked receiver." self stackValue: argumentCount + 1 "ret addr" put: localAbsentReceiver. (self maybeMethodHasCogMethod: newMethod) ifTrue: [ cogMethod := self cogMethodOf: newMethod. cogMethod selector = objectMemory nilObject ifTrue: [cogit setSelectorOf: cogMethod to: messageSelector] ifFalse: ["Deal with anonymous accessors, e.g. in Newspeak. The cogMethod may not have the correct selector. If not, try and compile a new method with the correct selector." cogMethod selector ~= messageSelector ifTrue: [ (cogit cog: newMethod selector: messageSelector) ifNotNil: [:newCogMethod | cogMethod := newCogMethod]]]. cogMethod selector = messageSelector ifTrue: [cogit linkNSSendCache: nsSendCache classTag: methodReceiverClassTag enclosingObject: localAbsentReceiverOrZero target: cogMethod caller: self mframeHomeMethodExport] ifFalse: ["Out of code memory. Fall through to interpret."]. instructionPointer := self popStack. self executeNewMethod. self unreachable]. instructionPointer := self popStack. self interpretMethodFromMachineCode. self unreachable.!

1 0

[commit][3428] Make the installer builders use SqueakV50.sources.
by commits＠squeakvm.org 22 Aug '15

22 Aug '15

Revision: 3428 Author: eliot Date: 2015-08-22 09:40:42 -0700 (Sat, 22 Aug 2015) Log Message: ----------- Make the installer builders use SqueakV50.sources. Update the README info generated by uploadvms to mention Squeak 5. Modified Paths: -------------- branches/Cog/build.macos32x86/newspeak.cog.spur/installer/Makefile branches/Cog/build.macos32x86/newspeak.cog.spur/makevm branches/Cog/build.win32x86/newspeak.cog.spur/installer/Makefile branches/Cog/scripts/uploadvms Modified: branches/Cog/build.macos32x86/newspeak.cog.spur/installer/Makefile =================================================================== --- branches/Cog/build.macos32x86/newspeak.cog.spur/installer/Makefile 2015-08-22 03:39:38 UTC (rev 3427) +++ branches/Cog/build.macos32x86/newspeak.cog.spur/installer/Makefile 2015-08-22 16:40:42 UTC (rev 3428) @@ -5,7 +5,7 @@ # edit and run ./links to get .app and sources, then run make SOURCES := yes -SOURCEFILENAME1 := $(shell echo ~/Squeak/SqueakV41.sources) +SOURCEFILENAME1 := ../../../sources/SqueakV50.sources DOCUMENTATION_DIR := doc include branding.gmk @@ -21,7 +21,7 @@ $(VM_LOCALIZED_NAME_ESC).app: ../Fast.app rm -rf $(VM_LOCALIZED_NAME_ESC).app cp -R ../Fast.app $(VM_LOCALIZED_NAME_ESC).app - -ln $(SOURCEFILENAME1) $(VM_LOCALIZED_NAME_ESC).app/Contents/Resources + ln $(SOURCEFILENAME1) $(VM_LOCALIZED_NAME_ESC).app/Contents/Resources # Cab contents: VM_BUNDLE := $(VM_LOCALIZED_NAME_ESC).app Modified: branches/Cog/build.macos32x86/newspeak.cog.spur/makevm =================================================================== --- branches/Cog/build.macos32x86/newspeak.cog.spur/makevm 2015-08-22 03:39:38 UTC (rev 3427) +++ branches/Cog/build.macos32x86/newspeak.cog.spur/makevm 2015-08-22 16:40:42 UTC (rev 3428) @@ -93,4 +93,4 @@ # Clean out any .svn directories (result of copying) in the VM find "$DEST" -name .svn -exec rm -rf {} \; #link source file into resources directory -ln ../../sources/SqueakV41.sources "$DEST/Contents/Resources" +ln ../../sources/SqueakV50.sources "$DEST/Contents/Resources" Modified: branches/Cog/build.win32x86/newspeak.cog.spur/installer/Makefile =================================================================== --- branches/Cog/build.win32x86/newspeak.cog.spur/installer/Makefile 2015-08-22 03:39:38 UTC (rev 3427) +++ branches/Cog/build.win32x86/newspeak.cog.spur/installer/Makefile 2015-08-22 16:40:42 UTC (rev 3428) @@ -17,7 +17,7 @@ VM_EXECUTABLE_MANIFEST := $(VM_EXECUTABLE).manifest VM_MANIFEST_8DOT3 := $(VM_BASE_NAME).man VM_ICON := $(VM_BASE_NAME).ico -SOURCEFILENAME1 := SqueakV41.sources +SOURCEFILENAME1 := SqueakV50.sources CAB_FILES := $(VM_EXECUTABLE) $(VM_EXECUTABLE_MAP) $(VM_EXECUTABLE_MANIFEST) ifeq ($(strip $(SOURCES)),yes) Modified: branches/Cog/scripts/uploadvms =================================================================== --- branches/Cog/scripts/uploadvms 2015-08-22 03:39:38 UTC (rev 3427) +++ branches/Cog/scripts/uploadvms 2015-08-22 16:40:42 UTC (rev 3428) @@ -192,7 +192,7 @@ non-blocking FFI calls. The archives containing "Spur" or "spur" are VMs using the new Spur object representation and garbage collector and should be used with Spur-format Squeak/Pharo/Croquet or Newspeak images. - Newspeak has moved to Spur and no longer supports non-Spur versions. + Squeak V5 and Newspeak have moved to Spur. Linux There are two variants of the Linux VMs; those ending in "ht" have a

1 0

Re: [Vm-dev] VM Maker: VMMaker.oscog-eem.1441.mcz
by Chris Cunningham 22 Aug '15

22 Aug '15

Hi. On Fri, Aug 21, 2015 at 6:19 PM, <commits(a)source.squeak.org> wrote: > > Eliot Miranda uploaded a new version of VMMaker to project VM Maker: > http://source.squeak.org/VMMaker/VMMaker.oscog-eem.1441.mcz > > ==================== Summary ==================== > > Name: VMMaker.oscog-eem.1441 > Author: eem > Time: 21 August 2015, 6:18:57.387 pm > UUID: c41d605b-7d2e-4ecc-95e1-b295119106a7 > Ancestors: VMMaker.oscog-eem.1440 > > Modify Spur ImageSegment load to become the segmentWordArray into an Array > of the loaded objects if load is successful, hence decoupling ImageSegment > from the assumption that objects are allocated in order. > > Fix the (unused) ByteArray>>long64At: extension method to do what it > advertises. > thanks, nice and clear and the following works: (ByteArray new: 8) long64At: 1 put: 32; long64At: 1 "==> 32" -cbc

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev August 2015