Vm-dev March 2016

vm-dev@lists.squeakfoundation.org

33 participants
139 discussions

error "please insert disk" (windows vm)
by Nicolai Hess 14 Jan '17

14 Jan '17

we have the following bug report 16928 <https://pharo.fogbugz.com/f/cases/16928/no-disk-in-the-drive-reply-to-World…> "no disk in the drive" reply to World | Save As (and other places) "On a clean Pharo4.0 #40618 image, on Windows 7 World Menu | Save As brings up a modal alert window, "There is no disk in the drive. Please insert a disk into drive <etc>" On my system, it does this 4 times. I'm assuming that each instance refers to one of the card-reader drives which are attached, but without a card inserted. This is a pain to deal with every single time a File Browser or World | Save As dialog is opened. Could the alerts either be - only raised when that specific drive is requested by a user action or, less preferably - given by a non-modal window, and fading away after a period" I was able to reproduce this error on squeaks and pharos window vm. (insert a usb card reader, with card, wait some time , remove only the card, every access on this drive, for example open a FileList and scroll to the drive letter). The solution I found is to call SetErrorMode( *SEM_FAILCRITICALERRORS)* I would like to propose the following change to sqwin32directory.c wrap the call to FindFirstFileW/FindNextFileW with UINT prevMode = SetErrorMode(SEM_FAILCRITICALERRORS); call FindFirstFileW/FindNextFileW SetErrorMode(prevMode); ... and call this functions in at least dir_lookup() (the other file/directory methods are save). Alternative solution, we could call the SetErrorMode function once at program start but I don't know if there are other "useful" errors message that we don't want to disable. I tested both solution, both seems to work fine. What do you think?

1 3

ffi-test-main.c fails on FreeBSD
by Gary Dunn 12 Aug '16

12 Aug '16

In conjunction with getting Open Cobalt running on FreeBSD I have been looking into the problems with SqueakFFIPrims on FreeBSD and have reached a point that is beyond my understanding. I am not sure if the fix should come from the VM side or the FreeBSD port maintainer. I needed to get source code back into the ports tree so I did a make in /usr/ports/lang/squeak. I already had the 3.9 tarball, so that just repopulated the source and did the initial build. In the "files" folder I found patch files typical for a FreeBSD port. These patch source files from a generic tarball to work with FreeBSD. The one of interest was patch-platforms__unix__plugins__SqueakFFIPrims__ffi-config which contains ----x-----x----- --- platforms/unix/plugins/SqueakFFIPrims/ffi-config.org Wed Apr 26 20:27:53 2006 +++ platforms/unix/plugins/SqueakFFIPrims/ffi-config Wed Apr 26 20:29:00 2006 @@ -39,6 +39,7 @@ case ${abi} in linux) abi=sysv;; + freebsd*) abi=sysv;; darwin*) abi=darwin;; *) abi=libffi; lib="-lffi";; esac -----x-----x----- I then had the folder work/Squeak-3.9-7/platforms/unix/plugins/SqueakFFIPrims In it is a file 00README which explains how to test FFI. It says to run ffi-test-config but I have no such file. It says to type "make" to build a test suite but the build failed. It says to try with make CPU=any ABI=libffi LIB=-lffi which also failed. I traced the problem to the gcc search paths for includes and libraries -- turns out FreeBSD only searches /usr/include and /usr/lib when the required files are in /usr/local/include and /usr/local/lib. After some fiddling I finally had a "main" to test. It failed: ffi-test-main.c failed at line 361. Here is a bit from that file, ending with line 361: ffiInitialize(); for (ul= 0; ul < 15; ++ul) ffiPushSingleFloat(fa[ul]); GO(FFITypeSingleFloat, many); f= ffiReturnFloatValue(); ffiCleanup(); CHECK(f - ff < FLT_EPSILON); I worked out the CHECK macro but could not find what FLT_EPSILON is. At that point I decided to stop and come to you for help. Where do I go from here? -- Gary Dunn, Honolulu osp(a)aloha.com http://openslate.net/ http://e9erust.blogspot.com/ Sent from Slate001

5 8

Tasks of the heartbeat code
by Holger Freyther 27 Jul '16

27 Jul '16

Dear Eliot, I tried to understand the basic load of the VM and I have two questions for the heartbeat code. Is it right that there are two concerns for the code? 1.) Call forceInterruptCheckFromHeartbeat periodically 2.) Update the clock as seen by the image 1st is useful to interrupt Processes like […] repeat with a pre-emptive scheduler? On a different note do backwards jumps generate interrupt checks too? 2nd is this an optimization to not have to generate the clock when the primitive is being used? thanks holger

4 12

Tracing a Spur Image from Smalltalk
by timfelgentreff 03 Jun '16

03 Jun '16

I've updated the SystemTracer package to trace a Spur-format image. This works as far as I can tell, producing a header that looks just like the header produced by the VM and the objects written also look good. I can also open the image in RSqueak, but not on Spur. It crashes immediately and produces a Smalltalk backtrace that looks like it never manages to return from the image-saving method (but the names in the Stack-trace all look ok, leading me to believe that enough objects got written ok to figure out classes, method names, find the right Process to run and so forth). Curiously, the old SystemTracer2 class that writes Cog-format images also produces images that cannot be opened by Cog VMs, only Interpreter VMs. So I assume there are other assumptions that maybe the JIT makes when starting up. So my question is, what assumptions could there be, and where could I start looking. Cheers, Tim (If you're wondering why I'm using the Smalltalk-level tracing at all - in our quest to create (with RSqueakVM) a Squeak VM that runs Squeak code fast enough that we no longer have to rely on C code from plugins or optional primitives, we're trying to see how far we can push this, and e.g. write the image from within itself, too.) -- View this message in context: http://forum.world.st/Tracing-a-Spur-Image-from-Smalltalk-tp4881224.html Sent from the Squeak VM mailing list archive at Nabble.com.

4 11

Re: [Pharo-dev] Epicea tests failure under Windows (WAS: RE: [ANN] We are in "code freeze" for Pharo 5)
by Nicolai Hess 15 May '16

15 May '16

I traced this back to Object>>withArgs:executeMethod: (primitive 188) in OpalCompiler>>#evaluate It does not happen on squeaks spur vm as far as I know (but I don't know what is the latests squeak spur vm). 2016-02-08 5:17 GMT+01:00 Martin Dias <tinchodias(a)gmail.com>: > Very interesting... > > I created case 17543 > <https://pharo.fogbugz.com/f/cases/17543/VM-crash-in-Windows-when-compiling-…>, > with Vincent's script and other quotes from this thread. I did it because > this Windows issue is independent from Epicea (whose integration only > exposed the Windows issue). > > Martin > > On Sat, Feb 6, 2016 at 6:41 AM, Max Leske <maxleske(a)gmail.com> wrote: > >> Changing the scheduler didn’t have any effect. However, it seems to me >> that the size of the input matters. I can make it work in two ways: >> 1. delete a line. If the line’s too short it won’t work though >> 2. delete a couple of characters from some symbols >> >> That would suggest that not the number of symbols / items in the array is >> significant but the number of bytes. So maybe there’s a problem with memory >> allocation in the VM. I haven’t been able to come up with an exact number, >> but the VM starts crashing at around 7096 characters. Sometimes I can go >> over 7100, other times not. Smaller numbers always work. >> >> HTH, >> Max >> >> On 06 Feb 2016, at 07:25, Ben Coman <btc(a)openInWorld.com >> <btc(a)openinworld.com>> wrote: >> >> On Sat, Feb 6, 2016 at 12:59 PM, Martin Dias <tinchodias(a)gmail.com> >> wrote: >> >> In fact, this integration step >> (https://ci.inria.fr/pharo/job/Pharo-5.0-Update-Step-2.1-Validation-M-Z) >> was >> already passed by Epicea in December. That was before spur, it might be >> related to spur. >> >> That time, the integration of Epicea revealed a bug in delay's >> scheduling... >> >> >> Was that these issues integrated in build 50466... >> https://pharo.fogbugz.com/default.asp?17066 >> https://pharo.fogbugz.com/default.asp?13756 >> >> which should have left the current scheduler as >> DelayExperimentalSpinScheduler >> but in current build 50564 "Delay delaySchedulerClass" --> >> DelayMicrosecondScheduler >> >> Maybe Delay class >> initialize has been executed since >> Scheduler ifNotNil: [ Scheduler stopTimerEventLoop ]. >> Scheduler := DelayMicrosecondScheduler new. >> Scheduler startTimerEventLoop. >> Smalltalk addToStartUpList: self. >> >> but I don't see any updates that might have invoked it. >> Anyway, maybe changing that will help. >> >> cheers -ben >> >> >> >> maybe now it's revealing some bug in Windows+spur...? >> >> Martin >> >> >> On Fri, Feb 5, 2016 at 10:09 AM, Blondeau Vincent >> <vincent.blondeau(a)worldline.com> wrote: >> >> >> If you remove any line, it stops crashing… >> >> I really don’t know what is going on… >> >> >> >> Vincent >> >> >> >> De : Pharo-dev [mailto:pharo-dev-bounces@lists.pharo.org >> <pharo-dev-bounces(a)lists.pharo.org>] De la part de >> Henrik Nergaard >> Envoyé : vendredi 5 février 2016 14:03 >> >> >> À : Pharo Development List >> Objet : Re: [Pharo-dev] Epicea tests failure under Windows (WAS: RE: [ANN] >> We are in "code freeze" for Pharo 5) >> >> >> >> Removing >> >> #(IRPrinterV2 #(visitStoreTemp: visitStoreRemoteTemp: >> visitPopIntoLiteralVariable: visitPushTemp: visitReturnLiteral: >> visitStoreInstVar: visitStoreLiteralVariable: visitPushLiteralVariable: >> visitPushInstVar: visitJump: visitPushLiteral: label: visitPushArray: >> visitPopIntoTemp: visitReturnInstVar: visitPopIntoRemoteTemp: >> visitTempVector: visitJumpIf: visitPushRemoteTemp: visitPopIntoInstVar: >> visitSend: )) >> >> And it stops crashing. >> >> >> >> - BTW, there is not crash.dmp, does it works with spur? >> >> >> >> https://pharo.fogbugz.com/f/cases/17506/when-an-Image-crashes-a-crash-dmp-f… >> >> >> >> Best regards, >> >> Henrik >> >> >> >> From: Pharo-dev [mailto:pharo-dev-bounces@lists.pharo.org] On Behalf Of >> Blondeau Vincent >> Sent: Friday, February 5, 2016 1:54 PM >> To: Pharo Development List <pharo-dev(a)lists.pharo.org> >> Subject: Re: [Pharo-dev] Epicea tests failure under Windows (WAS: RE: >> [ANN] We are in "code freeze" for Pharo 5) >> >> >> >> Very strange bug indeed that happens on windows machines that crashes the >> VM. I succeeded to reduce the problem to an evaluation of a collection… >> >> Can someone have the same issue under Windows or Mac? >> >> >> >> To reproduce : Copy paste the contents of the attached file in the >> playground and print it or do it >> >> It works on pre-spur vm. >> >> >> >> BTW, there is not crash.dmp, does it works with spur? >> >> Vincent >> >> >> >> De : Pharo-dev [mailto:pharo-dev-bounces@lists.pharo.org] De la part de >> Marcus Denker >> Envoyé : vendredi 5 février 2016 13:01 >> À : Pharo Development List >> Objet : Re: [Pharo-dev] Epicea tests failure under Windows (WAS: RE: [ANN] >> We are in "code freeze" for Pharo 5) >> >> >> >> >> >> On 05 Feb 2016, at 12:40, Marcus Denker <marcus.denker(a)inria.fr> wrote: >> >> >> >> I will try do integrate it now… >> >> >> >> >> >> >> >> No, it still fails. >> >> >> >> >> >> https://ci.inria.fr/pharo/job/Pharo-5.0-Update-Step-2.1-Validation-M-Z/labe… >> >> >> >> The last test it executed seems to be this: >> >> running suite: Refactoring-Tests-Environment >> >> starting testcase: RBBrowserEnvironmentTest>>testAndEnvironment ... >> >> Very strange... >> >> Marcus >> >> >> >> >> >> ________________________________ >> >> >> Ce message et les pièces jointes sont confidentiels et réservés à l'usage >> exclusif de ses destinataires. Il peut également être protégé par le >> secret >> professionnel. Si vous recevez ce message par erreur, merci d'en avertir >> immédiatement l'expéditeur et de le détruire. L'intégrité du message ne >> pouvant être assurée sur Internet, la responsabilité de Worldline ne >> pourra >> être recherchée quant au contenu de ce message. Bien que les meilleurs >> efforts soient faits pour maintenir cette transmission exempte de tout >> virus, l'expéditeur ne donne aucune garantie à cet égard et sa >> responsabilité ne saurait être recherchée pour tout dommage résultant d'un >> virus transmis. >> >> This e-mail and the documents attached are confidential and intended >> solely for the addressee; it may also be privileged. If you receive this >> e-mail in error, please notify the sender immediately and destroy it. As >> its >> integrity cannot be secured on the Internet, the Worldline liability >> cannot >> be triggered for the message content. Although the sender endeavours to >> maintain a computer virus-free network, the sender does not warrant that >> this transmission is virus-free and will not be liable for any damages >> resulting from any virus transmitted. >> >> >> ________________________________ >> >> Ce message et les pièces jointes sont confidentiels et réservés à l'usage >> exclusif de ses destinataires. Il peut également être protégé par le >> secret >> professionnel. Si vous recevez ce message par erreur, merci d'en avertir >> immédiatement l'expéditeur et de le détruire. L'intégrité du message ne >> pouvant être assurée sur Internet, la responsabilité de Worldline ne >> pourra >> être recherchée quant au contenu de ce message. Bien que les meilleurs >> efforts soient faits pour maintenir cette transmission exempte de tout >> virus, l'expéditeur ne donne aucune garantie à cet égard et sa >> responsabilité ne saurait être recherchée pour tout dommage résultant d'un >> virus transmis. >> >> This e-mail and the documents attached are confidential and intended >> solely for the addressee; it may also be privileged. If you receive this >> e-mail in error, please notify the sender immediately and destroy it. As >> its >> integrity cannot be secured on the Internet, the Worldline liability >> cannot >> be triggered for the message content. Although the sender endeavours to >> maintain a computer virus-free network, the sender does not warrant that >> this transmission is virus-free and will not be liable for any damages >> resulting from any virus transmitted. >> >> >> >

1 1

Spur corrupts large images when saving
by Yuriy Tymchuk 03 May '16

03 May '16

Dear vm developers, I’ve encountered multiple times an issue when my images cannot be opened after saving (only on spur). I think that this happens when image is large, and the following script results in a corrupt image almost 100%: curl get.pharo.org/50+vm | bash ./pharo Pharo.image --no-default-preferences eval --save \ "Smalltalk globals at: #ReallyBigArray put: (ByteArray new: 1024*1024*1000). 'Done'" ./pharo Pharo.image printVersion The last line is there just to make image do something and for me it fails all the times. I hope that this can help to troubleshoot the issue with the vm. Also here are my system & hardware specs: System Version: OS X 10.11.3 (15D21) Model Identifier: MacBookPro11,5 Processor Name: Intel Core i7 Processor Speed: 2.8 GHz Total Number of Cores: 4 Memory: 16 GB Cheers! Uko

15 57

[commit][3665] Squeak is actually not DPI aware (yet?).
by commits＠squeakvm.org 02 May '16

02 May '16

Revision: 3665 Author: topa Date: 2016-03-30 00:32:09 -0700 (Wed, 30 Mar 2016) Log Message: ----------- Squeak is actually not DPI aware (yet?). Modified Paths: -------------- branches/Cog/build.win32x86/newspeak.cog.spur/nsvm.exe.manifest branches/Cog/build.win32x86/newspeak.stack.spur/nsvm.exe.manifest branches/Cog/build.win32x86/squeak.cog.spur/Croquet.exe.manifest branches/Cog/build.win32x86/squeak.cog.spur/Squeak.exe.manifest branches/Cog/build.win32x86/squeak.cog.v3/Croquet.exe.manifest branches/Cog/build.win32x86/squeak.cog.v3/Squeak.exe.manifest branches/Cog/build.win32x86/squeak.stack.spur/Croquet.exe.manifest branches/Cog/build.win32x86/squeak.stack.spur/Squeak.exe.manifest branches/Cog/build.win32x86/squeak.stack.v3/Croquet.exe.manifest branches/Cog/build.win32x86/squeak.stack.v3/Squeak.exe.manifest Modified: branches/Cog/build.win32x86/newspeak.cog.spur/nsvm.exe.manifest =================================================================== --- branches/Cog/build.win32x86/newspeak.cog.spur/nsvm.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/newspeak.cog.spur/nsvm.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Newspeak Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/newspeak.stack.spur/nsvm.exe.manifest =================================================================== --- branches/Cog/build.win32x86/newspeak.stack.spur/nsvm.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/newspeak.stack.spur/nsvm.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Newspeak Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/squeak.cog.spur/Croquet.exe.manifest =================================================================== --- branches/Cog/build.win32x86/squeak.cog.spur/Croquet.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/squeak.cog.spur/Croquet.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Croquet Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/squeak.cog.spur/Squeak.exe.manifest =================================================================== --- branches/Cog/build.win32x86/squeak.cog.spur/Squeak.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/squeak.cog.spur/Squeak.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Squeak Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/squeak.cog.v3/Croquet.exe.manifest =================================================================== --- branches/Cog/build.win32x86/squeak.cog.v3/Croquet.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/squeak.cog.v3/Croquet.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Croquet Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/squeak.cog.v3/Squeak.exe.manifest =================================================================== --- branches/Cog/build.win32x86/squeak.cog.v3/Squeak.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/squeak.cog.v3/Squeak.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Squeak Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/squeak.stack.spur/Croquet.exe.manifest =================================================================== --- branches/Cog/build.win32x86/squeak.stack.spur/Croquet.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/squeak.stack.spur/Croquet.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Croquet Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/squeak.stack.spur/Squeak.exe.manifest =================================================================== --- branches/Cog/build.win32x86/squeak.stack.spur/Squeak.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/squeak.stack.spur/Squeak.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Squeak Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/squeak.stack.v3/Croquet.exe.manifest =================================================================== --- branches/Cog/build.win32x86/squeak.stack.v3/Croquet.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/squeak.stack.v3/Croquet.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Croquet Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency> Modified: branches/Cog/build.win32x86/squeak.stack.v3/Squeak.exe.manifest =================================================================== --- branches/Cog/build.win32x86/squeak.stack.v3/Squeak.exe.manifest 2016-03-29 20:18:32 UTC (rev 3664) +++ branches/Cog/build.win32x86/squeak.stack.v3/Squeak.exe.manifest 2016-03-30 07:32:09 UTC (rev 3665) @@ -12,7 +12,7 @@ <description>Squeak Smalltalk Virtual Machine</description> <asmv3:application> <asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> - <dpiAware>true</dpiAware> + <dpiAware>false</dpiAware> </asmv3:windowsSettings> </asmv3:application> <dependency>

5 6

OwnedLock primitives request for review
by Ben Coman 12 Apr '16

12 Apr '16

I have the OwnedLock primitives and image side code suggested by Dennis for Pharo ready for review... (207 #primitiveOwnedLockRelease) (208 primitiveOwnedLockTryAcquire) (209 primitiveOwnedLockWaitAcquire) Dennis identified an interesting corner case codified in testWaitingProcessReceivingLockTerminatedBeforeResuming. 1) low priority process B waits on lock because it is owned by another process A 2) process A release lock but process B not resumed since higher priority process C is running 3) process C terminates process B 4) process B should (but does not) release acquired lock during termination. This is because B is given the lock by A's call to primitivexxRelease, and the ensure in #critical doesn't wrap B's call to primitivexxWaitAcquire. However wrapping an ensure around the whole #critical means if the process is terminated while waiting, you need to be careful not to releasing someone else's lock. All a little bit messy. Now I read recently something about primitive handling of forwarders being done by side-effect-free failing of primitives and retying, and I wonder if the same might be feasible here. Rather than A's call to primitivexxRelease setting B as the lock owner, it just sets the owner to nil, and when B is woken up, it retries primitivexxWaitAcquire. How does that sound? cheers -ben

1 3

Re: [Vm-dev] VM Maker: VMMaker.oscog-eem.1755.mcz
by Eliot Miranda 31 Mar '16

31 Mar '16

On Thu, Mar 31, 2016 at 2:27 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.1755.mcz > > ==================== Summary ==================== > > Name: VMMaker.oscog-eem.1755 > Author: eem > Time: 1 January 1970, 2:26:33.237515 pm > ^^Ouch^^ !! That's what one gets for living on the bleeding edge :-/ > UUID: adcfc67f-a856-48b8-aef2-71ff6275a903 > Ancestors: VMMaker.oscog-nice.1754 > > Spur Cogit: > Add a compile-time flag, CheckRememberedInTrampoline, that controls > whether the remembered check is generated in-line or in the trampoline. If > checked in the trampoline and on 64-bits use one memory access to get the > header and share the access between the IMMUTABILITY check and the > rememberedBit check. Factor out the rememberedBit check into its own > method. > > The default leaves the code unchanged. I doubt that the option will > improve anything significantly; measuring the generated code in the > simulator the option saves only 0.6% of generated code. But it's so lovely > being able to explore the question in a couple of hours instead of at least > a day or two in C. > > Here's the data: CRIT=CheckRememberedInTrampoline; this methid is the last > method generated in a listener image before it prompts the user: > IMMTABILITY: 16r93F80 <-> 16r940E8: method: 16rC582F8 prim 117 > selector: 16r6D12C8 primRead:into:startingAt:count: > IMMTABILICRIT: 16r93250 <-> 16r933B8: method: 16rC582F8 prim 117 > selector: 16r6D12C8 primRead:into:startingAt:count: > VANILLA: 16r91C00 <-> 16r91D68: method: 16rC582F8 prim > 117 selector: 16r6D12C8 primRead:into:startingAt:count: > VANILLACRIT: 16r90D80 <-> 16r90EE8: method: 16rC582F8 prim 117 > selector: 16r6D12C8 primRead:into:startingAt:count: > 16r93F80 - 16r93250 / 16r93250 * 100.0 0.5601422920704027 > 16r91C00 - 16r90D80 / 16r90D80 * 100.0 0.6256742179072277 > > And interestingly IMMUTABILITY adds only 1.5% to the code bulk; writes are > rare. > 16r93F80 - 16r91C00 / 16r91C00 * 100.0 1.5222984562607205 > 16r93250 - 16r90D80 / 16r90D80 * 100.0 1.5884573894282634 > > =============== Diff against VMMaker.oscog-nice.1754 =============== > > Item was added: > + ----- Method: CogObjectRepresentation class>>initializeMiscConstants (in > category 'initialization') ----- > + initializeMiscConstants > + "Override to avoid inheriting (and hence repeating) VMClass > class>>initializeMiscConstants. > + Subclasses that have misc constants to initialize will override > further."! > > Item was changed: > CogObjectRepresentation subclass: #CogObjectRepresentationForSpur > instanceVariableNames: 'ceScheduleScavengeTrampoline > ceSmallActiveContextInMethodTrampoline > ceSmallActiveContextInBlockTrampoline > ceLargeActiveContextInMethodTrampoline > ceLargeActiveContextInBlockTrampoline ceStoreCheckContextReceiverTrampoline > ceStoreTrampolines' > + classVariableNames: 'CheckRememberedInTrampoline > NumStoreTrampolines' > - classVariableNames: 'NumStoreTrampolines' > poolDictionaries: 'VMBytecodeConstants VMSqueakClassIndices' > category: 'VMMaker-JIT'! > > Item was added: > + ----- Method: CogObjectRepresentationForSpur > class>>initializeMiscConstants (in category 'initialization') ----- > + initializeMiscConstants > + CheckRememberedInTrampoline := initializationOptions at: > #CheckRememberedInTrampoline ifAbsent: [false]! > > Item was added: > + ----- Method: > CogObjectRepresentationForSpur>>genCheckRememberedBitOf:scratch: (in > category 'compile abstract instructions') ----- > + genCheckRememberedBitOf: objReg scratch: scratchReg > + "Check the remembered bit of the object in objReg; answer the jump > taken if the bit is already set. > + Only need to fetch the byte containing it, which reduces the size > of the mask constant." > + | rememberedBitByteOffset mask | > + rememberedBitByteOffset := cogit backEnd isBigEndian > + > ifTrue: [objectMemory baseHeaderSize - 1 - (objectMemory > rememberedBitShift // 8)] > + > ifFalse:[objectMemory rememberedBitShift // 8]. > + mask := 1 << (objectMemory rememberedBitShift \\ 8). > + cogit MoveMb: rememberedBitByteOffset r: objReg R: scratchReg. > + cogit TstCq: mask R: scratchReg. > + ^cogit JumpNonZero: 0! > > Item was changed: > ----- Method: > CogObjectRepresentationForSpur>>genStoreCheckReceiverReg:valueReg:scratchReg:inFrame: > (in category 'compile abstract instructions') ----- > genStoreCheckReceiverReg: destReg valueReg: valueReg scratchReg: > scratchReg inFrame: inFrame > "Generate the code for a store check of valueReg into destReg." > + | jmpImmediate jmpDestYoung jmpSourceOld jmpAlreadyRemembered | > - | jmpImmediate jmpDestYoung jmpSourceOld jmpAlreadyRemembered mask > rememberedBitByteOffset | > <var: #jmpImmediate type: #'AbstractInstruction *'> > <var: #jmpDestYoung type: #'AbstractInstruction *'> > <var: #jmpSourceOld type: #'AbstractInstruction *'> > <var: #jmpAlreadyRemembered type: #'AbstractInstruction *'> > "Is value stored an immediate? If so we're done" > jmpImmediate := self genJumpImmediate: valueReg. > "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: valueReg. "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." > + CheckRememberedInTrampoline ifFalse: > + [jmpAlreadyRemembered := self genCheckRememberedBitOf: > destReg scratch: scratchReg]. > - 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. > "Remembered bit is not set. Call store check to insert dest into > remembered table." > self assert: destReg = ReceiverResultReg. > cogit > evaluateTrampolineCallBlock: [cogit CallRT: > ceStoreCheckTrampoline] > protectLinkRegIfNot: inFrame. > jmpImmediate jmpTarget: > (jmpDestYoung jmpTarget: > (jmpSourceOld jmpTarget: > + cogit Label)). > + CheckRememberedInTrampoline ifFalse: > + [jmpAlreadyRemembered jmpTarget: jmpSourceOld > getJmpTarget]. > - (jmpAlreadyRemembered jmpTarget: > - cogit Label))). > ^0! > > Item was added: > + ----- Method: CogObjectRepresentationForSpur>>genStoreCheckTrampoline > (in category 'initialization') ----- > + genStoreCheckTrampoline > + | jumpSC | > + <var: #jumpSC type: #'AbstractInstruction *'> > + <inline: true> > + CheckRememberedInTrampoline ifTrue: > + [cogit zeroOpcodeIndex. > + jumpSC := self genCheckRememberedBitOf: ReceiverResultReg > scratch: cogit backEnd cResultRegister. > + self assert: jumpSC opcode = JumpNonZero. > + jumpSC opcode: JumpZero. > + cogit RetN: 0. > + jumpSC jmpTarget: cogit Label]. > + ^cogit > + genTrampolineFor: #remember: > + called: 'ceStoreCheckTrampoline' > + numArgs: 1 > + arg: ReceiverResultReg > + arg: nil > + arg: nil > + arg: nil > + regsToSave: (cogit callerSavedRegMask bitClear: (cogit > registerMaskFor: ReceiverResultReg)) > + pushLinkReg: true > + resultReg: cogit returnRegForStoreCheck > + appendOpcodes: CheckRememberedInTrampoline! > > 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 jumpRC | > - | jumpSC | > <option: #IMMUTABILITY> > <var: #trampolineName type: #'char *'> > <var: #jumpSC type: #'AbstractInstruction *'> > + <var: #jumpRC 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. > cogit > compileTrampolineFor: > #ceCannotAssignTo:withIndex:valueToAssign: > numArgs: 3 > arg: ReceiverResultReg > arg: (instVarIndex < (NumStoreTrampolines - 1) > ifTrue: [cogit trampolineArgConstant: > instVarIndex] > ifFalse: [TempReg]) > arg: ClassReg > arg: nil > regsToSave: cogit emptyRegisterMask > pushLinkReg: true > resultReg: NoReg. > > "Store check" > jumpSC jmpTarget: cogit Label. > + "If on 64-bits and doing the remembered bit test here, we can > combine the tests to fetch the header once." > + CheckRememberedInTrampoline ifTrue: > + [objectMemory wordSize = 8 > + ifTrue: > + [cogit TstCq: 1 << objectMemory > rememberedBitShift R: SendNumArgsReg. > + jumpRC := cogit JumpZero: 0. > + cogit RetN: 0] > + ifFalse: > + [jumpRC := self genCheckRememberedBitOf: > ReceiverResultReg scratch: SendNumArgsReg. > + self assert: jumpRC opcode = JumpNonZero. > + jumpRC opcode: JumpZero. > + cogit RetN: 0]. > + jumpRC 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 > jmpAlreadyRemembered | > - | 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." > + CheckRememberedInTrampoline ifFalse: > + [jmpAlreadyRemembered := self genCheckRememberedBitOf: > destReg scratch: scratchReg]. > - 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. > needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. > > jmpImmediate jmpTarget: > (jmpDestYoung jmpTarget: > (jmpSourceOld jmpTarget: > + cogit Label)). > + CheckRememberedInTrampoline ifFalse: > + [jmpAlreadyRemembered jmpTarget: jmpSourceOld > getJmpTarget]. > - (jmpAlreadyRemembered 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: > [self cCode: [] inSmalltalk: > [ceStoreTrampolines := CArrayAccessor on: > (Array new: NumStoreTrampolines)]. > 0 to: NumStoreTrampolines - 1 do: > [:instVarIndex | > ceStoreTrampolines > at: instVarIndex > put: (self > > genStoreTrampolineCalled: (cogit > > trampolineName: > 'ceStoreTrampoline' > > numArgs: instVarIndex > > limit: NumStoreTrampolines - > 2) > instVarIndex: > instVarIndex)]]. > + ceStoreCheckTrampoline := self genStoreCheckTrampoline. > - 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: Cogit class>>initializeMiscConstants (in category 'class > initialization') ----- > initializeMiscConstants > super initializeMiscConstants. > Debug := initializationOptions at: #Debug ifAbsent: [false]. > (initializationOptions includesKey: #EagerInstructionDecoration) > ifTrue: > [EagerInstructionDecoration := > initializationOptions at: #EagerInstructionDecoration] > ifFalse: > [EagerInstructionDecoration ifNil: > [EagerInstructionDecoration := false]]. > "speeds up single stepping but could lose fidelity" > > ProcessorClass := (initializationOptions at: #ISA ifAbsentPut: > [self objectMemoryClass defaultISA]) caseOf: { > [#X64] > -> [BochsX64Alien]. > [#IA32] > -> [BochsIA32Alien]. > [#ARMv5] > -> [GdbARMAlien]. > [#MIPSEL] > -> [MIPSELSimulator] }. > CogCompilerClass := self activeCompilerClass. > (CogCompilerClass withAllSuperclasses copyUpTo: > CogAbstractInstruction) reverseDo: > [:compilerClass| compilerClass initialize; > initializeAbstractRegisters]. > + self objectMemoryClass objectRepresentationClass > initializeMiscConstants. > "Our criterion for which methods to JIT is literal count. The > default value is 60 literals or less." > MaxLiteralCountForCompile := initializationOptions at: > #MaxLiteralCountForCompile ifAbsent: [60]. > "we special-case 0, 1 & 2 argument sends, N is numArgs >= 3" > NumSendTrampolines := 4. > "Currently not even the ceImplicitReceiverTrampoline contains > object references." > NumObjRefsInRuntime := 0. > > NSCSelectorIndex := (NSSendCache instVarNames indexOf: #selector) > - 1. > NSCNumArgsIndex := (NSSendCache instVarNames indexOf: #numArgs) - > 1. > NSCClassTagIndex := (NSSendCache instVarNames indexOf: #classTag) > - 1. > NSCEnclosingObjectIndex := (NSSendCache instVarNames indexOf: > #enclosingObject) - 1. > NSCTargetIndex := (NSSendCache instVarNames indexOf: #target) - 1. > NumOopsPerNSC := NSSendCache instVarNames size. > > "Max size to alloca when compiling. > Mac OS X 10.6.8 segfaults approaching 8Mb. > Linux 2.6.9 segfaults above 11Mb. > WIndows XP segfaults approaching 2Mb." > MaxStackAllocSize := 1024 * 1024 * 3 / 2 ! > > Item was changed: > ----- Method: VMBasicConstants class>>namesDefinedAtCompileTime (in > category 'C translation') ----- > namesDefinedAtCompileTime > "Answer the set of names for variables that should be defined at > compile time. > Some of these get default values during simulation, and hence get > defaulted in > the various initializeMiscConstants methods. But that they have > values should > /not/ cause the code generator to do dead code elimination based > on their > default values." > ^#( VMBIGENDIAN > IMMUTABILITY > STACKVM COGVM COGMTVM SPURVM > + PharoVM > "Pharo vs Squeak" > + EnforceAccessControl > "Newspeak" > + CheckRememberedInTrampoline) "IMMUTABILITY"! > - "Pharo vs Squeak" PharoVM > - "Newspeak" EnforceAccessControl)! > > -- _,,,^..^,,,_ best, Eliot

1 0

VM Maker: VMMaker.oscog-eem.1755.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.1755.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.1755 Author: eem Time: 1 January 1970, 2:26:33.237515 pm UUID: adcfc67f-a856-48b8-aef2-71ff6275a903 Ancestors: VMMaker.oscog-nice.1754 Spur Cogit: Add a compile-time flag, CheckRememberedInTrampoline, that controls whether the remembered check is generated in-line or in the trampoline. If checked in the trampoline and on 64-bits use one memory access to get the header and share the access between the IMMUTABILITY check and the rememberedBit check. Factor out the rememberedBit check into its own method. The default leaves the code unchanged. I doubt that the option will improve anything significantly; measuring the generated code in the simulator the option saves only 0.6% of generated code. But it's so lovely being able to explore the question in a couple of hours instead of at least a day or two in C. Here's the data: CRIT=CheckRememberedInTrampoline; this methid is the last method generated in a listener image before it prompts the user: IMMTABILITY: 16r93F80 <-> 16r940E8: method: 16rC582F8 prim 117 selector: 16r6D12C8 primRead:into:startingAt:count: IMMTABILICRIT: 16r93250 <-> 16r933B8: method: 16rC582F8 prim 117 selector: 16r6D12C8 primRead:into:startingAt:count: VANILLA: 16r91C00 <-> 16r91D68: method: 16rC582F8 prim 117 selector: 16r6D12C8 primRead:into:startingAt:count: VANILLACRIT: 16r90D80 <-> 16r90EE8: method: 16rC582F8 prim 117 selector: 16r6D12C8 primRead:into:startingAt:count: 16r93F80 - 16r93250 / 16r93250 * 100.0 0.5601422920704027 16r91C00 - 16r90D80 / 16r90D80 * 100.0 0.6256742179072277 And interestingly IMMUTABILITY adds only 1.5% to the code bulk; writes are rare. 16r93F80 - 16r91C00 / 16r91C00 * 100.0 1.5222984562607205 16r93250 - 16r90D80 / 16r90D80 * 100.0 1.5884573894282634 =============== Diff against VMMaker.oscog-nice.1754 =============== Item was added: + ----- Method: CogObjectRepresentation class>>initializeMiscConstants (in category 'initialization') ----- + initializeMiscConstants + "Override to avoid inheriting (and hence repeating) VMClass class>>initializeMiscConstants. + Subclasses that have misc constants to initialize will override further."! Item was changed: CogObjectRepresentation subclass: #CogObjectRepresentationForSpur instanceVariableNames: 'ceScheduleScavengeTrampoline ceSmallActiveContextInMethodTrampoline ceSmallActiveContextInBlockTrampoline ceLargeActiveContextInMethodTrampoline ceLargeActiveContextInBlockTrampoline ceStoreCheckContextReceiverTrampoline ceStoreTrampolines' + classVariableNames: 'CheckRememberedInTrampoline NumStoreTrampolines' - classVariableNames: 'NumStoreTrampolines' poolDictionaries: 'VMBytecodeConstants VMSqueakClassIndices' category: 'VMMaker-JIT'! Item was added: + ----- Method: CogObjectRepresentationForSpur class>>initializeMiscConstants (in category 'initialization') ----- + initializeMiscConstants + CheckRememberedInTrampoline := initializationOptions at: #CheckRememberedInTrampoline ifAbsent: [false]! Item was added: + ----- Method: CogObjectRepresentationForSpur>>genCheckRememberedBitOf:scratch: (in category 'compile abstract instructions') ----- + genCheckRememberedBitOf: objReg scratch: scratchReg + "Check the remembered bit of the object in objReg; answer the jump taken if the bit is already set. + Only need to fetch the byte containing it, which reduces the size of the mask constant." + | rememberedBitByteOffset mask | + rememberedBitByteOffset := cogit backEnd isBigEndian + ifTrue: [objectMemory baseHeaderSize - 1 - (objectMemory rememberedBitShift // 8)] + ifFalse:[objectMemory rememberedBitShift // 8]. + mask := 1 << (objectMemory rememberedBitShift \\ 8). + cogit MoveMb: rememberedBitByteOffset r: objReg R: scratchReg. + cogit TstCq: mask R: scratchReg. + ^cogit JumpNonZero: 0! Item was changed: ----- Method: CogObjectRepresentationForSpur>>genStoreCheckReceiverReg:valueReg:scratchReg:inFrame: (in category 'compile abstract instructions') ----- genStoreCheckReceiverReg: destReg valueReg: valueReg scratchReg: scratchReg inFrame: inFrame "Generate the code for a store check of valueReg into destReg." + | jmpImmediate jmpDestYoung jmpSourceOld jmpAlreadyRemembered | - | jmpImmediate jmpDestYoung jmpSourceOld jmpAlreadyRemembered mask rememberedBitByteOffset | <var: #jmpImmediate type: #'AbstractInstruction *'> <var: #jmpDestYoung type: #'AbstractInstruction *'> <var: #jmpSourceOld type: #'AbstractInstruction *'> <var: #jmpAlreadyRemembered type: #'AbstractInstruction *'> "Is value stored an immediate? If so we're done" jmpImmediate := self genJumpImmediate: valueReg. "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: valueReg. "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." + CheckRememberedInTrampoline ifFalse: + [jmpAlreadyRemembered := self genCheckRememberedBitOf: destReg scratch: scratchReg]. - 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. "Remembered bit is not set. Call store check to insert dest into remembered table." self assert: destReg = ReceiverResultReg. cogit evaluateTrampolineCallBlock: [cogit CallRT: ceStoreCheckTrampoline] protectLinkRegIfNot: inFrame. jmpImmediate jmpTarget: (jmpDestYoung jmpTarget: (jmpSourceOld jmpTarget: + cogit Label)). + CheckRememberedInTrampoline ifFalse: + [jmpAlreadyRemembered jmpTarget: jmpSourceOld getJmpTarget]. - (jmpAlreadyRemembered jmpTarget: - cogit Label))). ^0! Item was added: + ----- Method: CogObjectRepresentationForSpur>>genStoreCheckTrampoline (in category 'initialization') ----- + genStoreCheckTrampoline + | jumpSC | + <var: #jumpSC type: #'AbstractInstruction *'> + <inline: true> + CheckRememberedInTrampoline ifTrue: + [cogit zeroOpcodeIndex. + jumpSC := self genCheckRememberedBitOf: ReceiverResultReg scratch: cogit backEnd cResultRegister. + self assert: jumpSC opcode = JumpNonZero. + jumpSC opcode: JumpZero. + cogit RetN: 0. + jumpSC jmpTarget: cogit Label]. + ^cogit + genTrampolineFor: #remember: + called: 'ceStoreCheckTrampoline' + numArgs: 1 + arg: ReceiverResultReg + arg: nil + arg: nil + arg: nil + regsToSave: (cogit callerSavedRegMask bitClear: (cogit registerMaskFor: ReceiverResultReg)) + pushLinkReg: true + resultReg: cogit returnRegForStoreCheck + appendOpcodes: CheckRememberedInTrampoline! 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 jumpRC | - | jumpSC | <option: #IMMUTABILITY> <var: #trampolineName type: #'char *'> <var: #jumpSC type: #'AbstractInstruction *'> + <var: #jumpRC 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. cogit compileTrampolineFor: #ceCannotAssignTo:withIndex:valueToAssign: numArgs: 3 arg: ReceiverResultReg arg: (instVarIndex < (NumStoreTrampolines - 1) ifTrue: [cogit trampolineArgConstant: instVarIndex] ifFalse: [TempReg]) arg: ClassReg arg: nil regsToSave: cogit emptyRegisterMask pushLinkReg: true resultReg: NoReg. "Store check" jumpSC jmpTarget: cogit Label. + "If on 64-bits and doing the remembered bit test here, we can combine the tests to fetch the header once." + CheckRememberedInTrampoline ifTrue: + [objectMemory wordSize = 8 + ifTrue: + [cogit TstCq: 1 << objectMemory rememberedBitShift R: SendNumArgsReg. + jumpRC := cogit JumpZero: 0. + cogit RetN: 0] + ifFalse: + [jumpRC := self genCheckRememberedBitOf: ReceiverResultReg scratch: SendNumArgsReg. + self assert: jumpRC opcode = JumpNonZero. + jumpRC opcode: JumpZero. + cogit RetN: 0]. + jumpRC 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 jmpAlreadyRemembered | - | 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." + CheckRememberedInTrampoline ifFalse: + [jmpAlreadyRemembered := self genCheckRememberedBitOf: destReg scratch: scratchReg]. - 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. needRestoreRcvr ifTrue: [ cogit putSelfInReceiverResultReg ]. jmpImmediate jmpTarget: (jmpDestYoung jmpTarget: (jmpSourceOld jmpTarget: + cogit Label)). + CheckRememberedInTrampoline ifFalse: + [jmpAlreadyRemembered jmpTarget: jmpSourceOld getJmpTarget]. - (jmpAlreadyRemembered 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: [self cCode: [] inSmalltalk: [ceStoreTrampolines := CArrayAccessor on: (Array new: NumStoreTrampolines)]. 0 to: NumStoreTrampolines - 1 do: [:instVarIndex | ceStoreTrampolines at: instVarIndex put: (self genStoreTrampolineCalled: (cogit trampolineName: 'ceStoreTrampoline' numArgs: instVarIndex limit: NumStoreTrampolines - 2) instVarIndex: instVarIndex)]]. + ceStoreCheckTrampoline := self genStoreCheckTrampoline. - 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: Cogit class>>initializeMiscConstants (in category 'class initialization') ----- initializeMiscConstants super initializeMiscConstants. Debug := initializationOptions at: #Debug ifAbsent: [false]. (initializationOptions includesKey: #EagerInstructionDecoration) ifTrue: [EagerInstructionDecoration := initializationOptions at: #EagerInstructionDecoration] ifFalse: [EagerInstructionDecoration ifNil: [EagerInstructionDecoration := false]]. "speeds up single stepping but could lose fidelity" ProcessorClass := (initializationOptions at: #ISA ifAbsentPut: [self objectMemoryClass defaultISA]) caseOf: { [#X64] -> [BochsX64Alien]. [#IA32] -> [BochsIA32Alien]. [#ARMv5] -> [GdbARMAlien]. [#MIPSEL] -> [MIPSELSimulator] }. CogCompilerClass := self activeCompilerClass. (CogCompilerClass withAllSuperclasses copyUpTo: CogAbstractInstruction) reverseDo: [:compilerClass| compilerClass initialize; initializeAbstractRegisters]. + self objectMemoryClass objectRepresentationClass initializeMiscConstants. "Our criterion for which methods to JIT is literal count. The default value is 60 literals or less." MaxLiteralCountForCompile := initializationOptions at: #MaxLiteralCountForCompile ifAbsent: [60]. "we special-case 0, 1 & 2 argument sends, N is numArgs >= 3" NumSendTrampolines := 4. "Currently not even the ceImplicitReceiverTrampoline contains object references." NumObjRefsInRuntime := 0. NSCSelectorIndex := (NSSendCache instVarNames indexOf: #selector) - 1. NSCNumArgsIndex := (NSSendCache instVarNames indexOf: #numArgs) - 1. NSCClassTagIndex := (NSSendCache instVarNames indexOf: #classTag) - 1. NSCEnclosingObjectIndex := (NSSendCache instVarNames indexOf: #enclosingObject) - 1. NSCTargetIndex := (NSSendCache instVarNames indexOf: #target) - 1. NumOopsPerNSC := NSSendCache instVarNames size. "Max size to alloca when compiling. Mac OS X 10.6.8 segfaults approaching 8Mb. Linux 2.6.9 segfaults above 11Mb. WIndows XP segfaults approaching 2Mb." MaxStackAllocSize := 1024 * 1024 * 3 / 2 ! Item was changed: ----- Method: VMBasicConstants class>>namesDefinedAtCompileTime (in category 'C translation') ----- namesDefinedAtCompileTime "Answer the set of names for variables that should be defined at compile time. Some of these get default values during simulation, and hence get defaulted in the various initializeMiscConstants methods. But that they have values should /not/ cause the code generator to do dead code elimination based on their default values." ^#( VMBIGENDIAN IMMUTABILITY STACKVM COGVM COGMTVM SPURVM + PharoVM "Pharo vs Squeak" + EnforceAccessControl "Newspeak" + CheckRememberedInTrampoline) "IMMUTABILITY"! - "Pharo vs Squeak" PharoVM - "Newspeak" EnforceAccessControl)!

1 0

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