Vm-dev July 2013

vm-dev@lists.squeakfoundation.org

20 participants
104 discussions

Issue 119 in cog: SerialPlugin primitiveOpenPortByName: ...
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: New Owner: ---- Labels: Type-Defect Priority-Medium New issue 119 by luc.fabr...(a)gmail.com: SerialPlugin primitiveOpenPortByName: ... http://code.google.com/p/cog/issues/detail?id=119 Since VMMaker-oscog-EstebanLorenzano.166, there is a bug in the implementation of SerialPlugin>>primitiveSerialPortOpenByName: portName baudRate: baudRate stopBitsType: stopBitsType parityType: parityType dataBits: dataBits inFlowControlType: inFlowControl outFlowControlType: outFlowControl xOnByte: xOnChar xOffByte: xOffChar Find a (small) patch attached. It is simple and it works. Just need integration. Luc Attachments: VMMaker-oscog-LucFabresse.229.mcz 2.6 MB

1 1

Issue 117 in cog: primitiveRemLargeIntegers is incorrect for negative quotient
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: Accepted Owner: nicolas....(a)gmail.com Labels: Type-Defect Priority-High New issue 117 by nicolas....(a)gmail.com: primitiveRemLargeIntegers is incorrect for negative quotient http://code.google.com/p/cog/issues/detail?id=117 Now that primitive 20 is connected (COG 2662), I had a chance to run some tests and discover a bug (stupid copy/paste from quo)... The result should be negated iff receiver negative. Attachments: InterpreterPrimitives-primitiveRemLargeIntegers.st 871 bytes

1 1

Issue 115 in cog: FFIPlugin unsignedShortAt: answer a signed short
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: Accepted Owner: nicolas....(a)gmail.com Labels: Type-Defect Priority-Medium New issue 115 by nicolas....(a)gmail.com: FFIPlugin unsignedShortAt: answer a signed short http://code.google.com/p/cog/issues/detail?id=115 (#[255 255 255 255] unsignedShortAt: 1) -> -1 This is exactly http://bugs.squeak.org/view.php?id=1415 It was reported in 2005, integrated in 2006 in VMMaker-tpr.55.mcz... ...and is still missing in COG branch Attachments: CogIssue115_FFIPlugin_unsignedShortAt_isSigned.st 1.6 KB

1 2

Issue 114 in cog: BMPReadWriterPlugin odes not check negative width and is subject to buffer overflow
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: Accepted Owner: nicolas....(a)gmail.com Labels: Type-Defect Priority-Medium New issue 114 by nicolas....(a)gmail.com: BMPReadWriterPlugin odes not check negative width and is subject to buffer overflow http://code.google.com/p/cog/issues/detail?id=114 This was reported and fixed by Andreas in 2006 at http://bugs.squeak.org/view.php?id=4360 And integrated in 2007 in VMMaker-tpr.64... I fail to see why it should not be in COG branch too. Attachments: CogIssue114_fix_a_BMPReadWriterPlugin_potentialBufferOverflow.st 3.4 KB

1 1

Issue 113 in cog: Fix divide-by-zero protection in JPEGReaderPlugin
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: Accepted Owner: nicolas....(a)gmail.com Labels: Type-Defect Priority-Medium New issue 113 by nicolas....(a)gmail.com: Fix divide-by-zero protection in JPEGReaderPlugin http://code.google.com/p/cog/issues/detail?id=113 Fix divideByZero protection in JPEGReaderPlugin as discussed in http://lists.squeakfoundation.org/pipermail/vm-dev/2012-December/011671.html Attachments: CogIssue113_JPEGReaderPlugin_divideByZero_protection.st 3.5 KB

1 1

Issue 112 in cog: Literal > 16rFFFFFFFF should better be generated as unsigned long long
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: Accepted Owner: nicolas....(a)gmail.com Labels: Type-Defect Priority-High New issue 112 by nicolas....(a)gmail.com: Literal > 16rFFFFFFFF should better be generated as unsigned long long http://code.google.com/p/cog/issues/detail?id=112 This was one change of issue 92 which was forgotten. Currently, gcc is kind enough to correct 0xffffffffffffffffUL by itself, but it issues a warning and that stinks like Undefined Behaviour. Find attached correction based on .oscog-emm.241

1 2

Issue 111 in cog: Use direction aware << and >> instead of bitShift: to avoid useless Runtime test
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: Accepted Owner: nicolas....(a)gmail.com Labels: Type-Enhancement Priority-Medium Performance Maintainability New issue 111 by nicolas....(a)gmail.com: Use direction aware << and >> instead of bitShift: to avoid useless Runtime test http://code.google.com/p/cog/issues/detail?id=111 As reported in vm-dev mailing list "bitShift: and runtime sign discussion" http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.vm.devel/9261 in (expr bitShift: shift), some shift have well known direction for the programmer, but the CCodeGenerator is unable to guess if shift is not literal. This results in useless runtime tests. Worse, it can increase the number of C compiler warnings after inlining. I attach some changes to VMMaker (based on .oscog-eem.241 branch). For sound, there is a shift sign discussion already, so we can eventually use directed shift too (based on trunk -ul.32 branch). Attachments: VMMaker_replace_bitShift_with_directedShift.cs 12.1 KB Sound_replace_bitShift_with_directedShift.cs 1.8 KB

1 3

Issue 108 in cog: Access to stackPages broken in Stack/CoInterpreter>>#commonReturn
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: New Owner: ---- Labels: Type-Defect Priority-Medium New issue 108 by stm...(a)gmail.com: Access to stackPages broken in Stack/CoInterpreter>>#commonReturn http://code.google.com/p/cog/issues/detail?id=108 One of the branches in Stack/CoInterpreter>>#commonReturn does a `self freeStackPage: thePage` where all other places in the code do `stackPages freeStackPage: thePage`. #freeStackPage: is not defined on self, so I suppose the attached patch is necessary. The bug should be in the CogVM branch and the Pharo branches of the codebase. Attachments: StackPages Access.1.cs 11.5 KB

1 1

Issue 107 in cog: Bit-rot in the Pharo CogVM code based (Simulator)
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: New Owner: ---- Labels: Type-Defect Priority-Medium New issue 107 by stm...(a)gmail.com: Bit-rot in the Pharo CogVM code based (Simulator) http://code.google.com/p/cog/issues/detail?id=107 The attached change set fixes bit rot in the *InterpreterSimulator and related classes. Here a brief summary: - VMClass>>#doOrDefer: now uses ProcessBrowser>>#isUIProcess: instead of Project - it replaces the use of #asDisplayText with StringMorph>>#contents:. - use of ThreadStream>>#on: is replaced with ThreadSafeTranscript>>#new - Utilities is replaced by UIManager>>#default - and #primitiveGetAttribute now uses the non-depricated `Smalltalk vm getSystemAttribute: attr` Attachments: Fix simulator.3.cs 30.0 KB

1 1

Issue 106 in cog: Bit-rot in NOMS>>#sqGrowMemory:By: and IS>>#ioUTCMicrosecondsNow
by cog＠googlecode.com 08 Aug '13

08 Aug '13

Status: New Owner: ---- Labels: Type-Defect Priority-Medium New issue 106 by stm...(a)gmail.com: Bit-rot in NOMS>>#sqGrowMemory:By: and IS>>#ioUTCMicrosecondsNow http://code.google.com/p/cog/issues/detail?id=106 In InterpreterSimulator the method #ioUTCMicrosecondsNow is missing. I just added this: ioUTCMicrosecondsNow "STEFAN: not entierly sure what to do with this... but the method is missing." ^ Time millisecondClockValue In the NewObjectMemorySimulator>>#sqGrowMemory:By:, there is a reference to coInterpreter which receives are #memory: message. However in the new hierarchy, where the interpreter is not an ObjectMemory anymore, that message is not implemented. So, I removed it... Not sure whether either of these solutions is appropriate, but would be good to get the problems fixed and have the simulators usable. Attachments: more-bit-rot.1.cs 857 bytes

1 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev July 2013