Vm-dev July 2009

vm-dev@lists.squeakfoundation.org

21 participants
20 discussions

Profiling and delay accuracy
by Andreas Raab 06 Aug '11

06 Aug '11

Hi Folks - For a long time now I had this feeling that MessageTally reports pretty bogus results in many situations without being able to precisely track down why this would be the case. Today I was looking at primitiveResponse() and, much to my surprise, noticed that the checkForInterrupts() that used to be there for years is gone (see Squeak3.6 for example which still has it). The sole reason for this check was to get both high delay accuracy and good profiling behavior. Consider this example: fc := FormCanvas on: Display. MessageTally spyOn:[ 1 to: 100 do:[:i| fc fillColor: Color white. fc printString. ]. ]. Using 3.6 this will be measured correctly as: **Tree** 87.5% {243ms} FormCanvas>>fillColor: |87.5% {243ms} GrafPort>>fillRect:offset: | 87.5% {243ms} GrafPort>>copyBits 12.5% {35ms} FormCanvas(Object)>>printString 12.5% {35ms} FormCanvas(Object)>>printStringLimitedTo: whereas the 3.10 reports it as **Tree** 99.6% {670ms} FormCanvas(Object)>>printString 94.9% {639ms} primitives 4.6% {31ms} FormCanvas(Object)>>printStringLimitedTo: I think you can see the problem and just imagine what this does to any cumulative results. I also wonder what that change does to platforms that don't have high-resolution asynchronous timers. This used to be the one place where we would catch timer changes at high resolution and unless the platform forces high resolution timer interrupts asynchronously (by resetting the interrupt counter) we must be loosing a lot in terms of delay accuracy. (it just so happens that the Windows VM *does* do asynchronous timer interrupts so I never noticed this but I'm very curious about the difference on Linux which typically still has pretty bad timer granularity). Any opinions? My feeling is that the timer check should go back in for all of the above reasons. At the very least a quickCheckForInterrupt() is required but I actually think putting the full check back in is worth it. Cheers, - Andreas

3 3

Joys of Squeak-dev image, VM Compiling on MacIntel and Spoon remote OtherBrowsing
by Ken G. Brown 16 Nov '10

16 Nov '10

MacPro OS X 10.4.9 Xcode 2.4.1 Squeak-dev-114-2, VMMaker 2.8b6.1, Squeak 3.8.15beta6U.app, and version 1657 of svn MacOS VM sources, and Spoon v2a12. Downloaded http://www.squeakvm.org/cgi-bin/viewcvs.cgi/ trunk version 1657 for MacOS Ran squeak-dev-beta-114-2.image on Squeak 3.8.15beta6U.app Loaded VMMaker 3.8b6.1 from dev image Universes. Generated vm sources by following instructions in <http://www.squeakvm.org/cgi-bin/viewcvs.cgi/trunk/platforms/Mac%20OS/vm/Doc…> Fought with Xcode wanting to use MacOSX10.3.9.sdk for ppc which was giving an error which was known and documented in the Readme. Finally figured out what Xcode wanted so built universal app just using 10.4.9 sdk. Couple issues: Exupery references remain in sqnamedPrims.h and sqPlatformSpecific.h from svn version 1657, I just commented them out. Notes said: Info-NPSqueak__Upgraded_.plist was to be found in the Developer directory. I could not find it so scammed a info.plist out of John's Squeak 3.8.15beta6U.app vm. Not sure what impact this has. And much, much diddling about figuring out various things... Finally got a build. It's a lot smaller (3.9MB) than John McIntosh's Squeak 3.8.15beta6U.app (8.1MB), and it looks like it is a PowerPC one but at least it runs. I'll look into the Xcode build process some more. Then: Replaced interp.c with Craig's interp.c for Spoon, rebuilt and got a VM with Spoon client stuff. Downloaded Spoon release 2a12. Moved a copy of Flow plugin from within Craig's naiad.app package into a plugins subdirectory where the new spoon client compatible MacIntel universal VM was run from. Followed Craig's Spoon instructions for loading niaid.st and niaid.pr into the 3.10 image (Squeak-dev-114-2) running on the new VM. Did the run spoon.app thing from Spoon.dmg (version 2a12), needs to be done twice if Safari is not running. Got Spoon processor running. Followed instructions again in the Spoon client and lo! Got OtherBrowser browsing the Spoon memory running on processor. Yay Craig! Who do i get to sign my certificate? :) Ken

4 7

Fwd: [Pharo-project] Could not load VMMaker in 10315
by John M McIntosh 18 Aug '09

18 Aug '09

Begin forwarded message: > From: Stéphane Ducasse <stephane.ducasse(a)inria.fr> > Date: May 24, 2009 1:35:01 AM PDT (CA) > To: Pharo Development <Pharo-project(a)lists.gforge.inria.fr> > Subject: [Pharo-project] Could not load VMMaker in 10315 > Reply-To: Pharo-project(a)lists.gforge.inria.fr > > Hi > > I did > > ScriptLoader loadFFI > Then loaded VMMaker-dtl.122 > Then proceeded when asked about klatt. > > And I got Cannot compile -- stack including temps is too deep > > what did I miss? > > Stef > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project -- = = = ======================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com = = = ========================================================================

4 7

Forwarded: [squeak-dev] Re: Problems building squeak VM (Fedora core 11)
by David T. Lewis 16 Aug '09

16 Aug '09

FYI for folks building VMs on Linux, the following is from the squeak-dev list. See the list archives for background. Dave ----- Forwarded message from Casimiro de Almeida Barreto <casimiro.barreto(a)gmail.com> ----- Date: Thu, 25 Jun 2009 00:16:06 -0300 From: Casimiro de Almeida Barreto <casimiro.barreto(a)gmail.com> To: The general-purpose Squeak developers list <squeak-dev(a)lists.squeakfoundation.org> Subject: [squeak-dev] Re: Problems building squeak VM (Fedora core 11) Precedence: list Problem is Fedora rel 11 compiler is botched. Made a copy of /usr/include/stdio.h in $ROOT/platforms/unix/vm/stdio.h Edited this file removing dprintf (starting line 399) Changed: #include <stdio.h> to: #include "stdio.h" whenever had problem. in file debug.c included line #include <stdlib.h> Dirty work. But fix things... :( ----- End forwarded message -----

7 13

Resources directory in SecurityPlugin
by Bert Freudenberg 04 Aug '09

04 Aug '09

When running Etoys inside a web browser, the file sandbox is preventing access to all files outside the "Etoys" working directory (a.k.a. "untrusted user directory"). That means we cannot open help guides or switch languages or access examples or the sources file etc. Proposed solution: In the SecurityPlugin, allow read-only file to a "resource" directory (in addition to read-write access to the "untrusted user directory"). The default for this resource directory would be the image directory. I did it for Unix, you can find a diff attached to the Etoys ticket: http://tracker.squeakland.org/browse/SQ-241 Basically, the writeFlag is now passed down to isAccessiblePathName(), and in case it's read-only, it allows access to the resource dir (which is configurable but defaulting to the image path). Any chance to get the same behavior implemented on Mac and Win VMs, too? Thanks :) - Bert -

2 2

patch: fix double alignment detection on Solaris/SPARC with gcc-4.4.x
by Andrew Gaylard 01 Aug '09

01 Aug '09

On Mon, Jul 20, 2009 at 7:02 PM, David T. Lewis<lewis(a)mail.msen.com> wrote: > > On Sat, Jul 18, 2009 at 05:25:48PM +0200, Andrew Gaylard wrote: >> >> The main issue that I'm facing at the moment is that gcc-4.4.0 generates >> code that causes unaligned accesses in fetchFloatAtinto; the same code >> works in gcc-4.2.4 on SPARC. Until I or someone else (hint!) fixes this, >> I'm continuing to build with gcc-4.2.4. > > fetchFloatAtInto() is implemented in platforms/Cross/vm/sqMemoryAccess.h > in the form of cpp macros. It might be tricky to debug this, but I note > that the implementation depends on DOUBLE_WORD_ALIGNMENT, which is a macro > that would have been set when you run configure. You might playing with > this setting to see if it gives you the results you want. For example, as > a complete SWAG, it might be the case that SPARC plus gcc-4.2.4 needs this > macro defined but for some reason configure is not setting it for you. David, thanks for the tip. This message from configure showed you are right (the 'yes' should be 'no'): checking whether unaligned access to doubles is ok... yes It turns out that newer GCCs, from 4.3.x onward, are smart enough to optimise out the code used to detect strict alignment: bash-3.2$ gcc -O2 double-test.c double-test.c: In function 'main': double-test.c:6: warning: passing argument 1 of 'f' makes integer from pointer without a cast double-test.c:1: note: expected 'int' but argument is of type 'char *' bash-3.2$ ./a.out bash-3.2$ echo $? 0 But when compiling without optimisation: bash-3.2$ gcc double-test.c double-test.c: In function 'main': double-test.c:6: warning: passing argument 1 of 'f' makes integer from pointer without a cast double-test.c:1: note: expected 'int' but argument is of type 'char *' bash-3.2$ ./a.out Bus Error (core dumped) ... as expected. This patch complicates the code sufficiently (by using the results of the function) so that it triggers a bus error both with and without optimisation, as would be expected. Of course, configure needs to be regenerated after applying this patch. Running our internal benchmark reveals that it is worth it getting 4.4.1 to work: gcc-4.4.1: real 61m45.975s user 56m16.247s sys 4m57.253s gcc-4.2.4: real 67m29.260s user 61m40.666s sys 5m7.937s This means the VM is sped up by 9%. - Andrew --- Squeak-3.10-4/platforms/unix/config/acinclude.m4.orig 2008-09-02 20:49:43.000000000 +0200 +++ Squeak-3.10-4/platforms/unix/config/acinclude.m4 2009-07-30 16:16:55.828929831 +0200 @@ -230,10 +230,33 @@ AC_DEFUN([AC_C_DOUBLE_ALIGNMENT], [AC_CACHE_CHECK([whether unaligned access to doubles is ok], ac_cv_double_align, - AC_TRY_RUN([f(int i){*(double *)i=*(double *)(i+4);} - int main(){char b[[12]];f(b);return 0;}], + AC_TRY_RUN([ + +int f(void* i){ + *(double *)i=*(double *)(i+4); + return *(char*)i; +} +int main(){ + char b[[12]]; + b[[0]]=1; + b[[1]]=2; + b[[2]]=3; + b[[3]]=4; + b[[4]]=0; + b[[5]]=0; + b[[6]]=0; + b[[7]]=0; + b[[8]]=0; + b[[9]]=0; + b[[10]]=0; + b[[11]]=0; + + return f(b); +} +], + ac_cv_double_align="yes", ac_cv_double_align="no")) test "$ac_cv_double_align" = "no" && AC_DEFINE(DOUBLE_WORD_ALIGNMENT)]) AC_DEFUN([AC_C_DOUBLE_ORDER],

3 5

Missing ShortIntegerArray coercions for use by SmartSyntaxPlugin (Was:Re: [Vm-dev] VMMaker: No senders of #ccg:emitLoadFor:from:on:)
by Igor Stasenko 29 Jul '09

29 Jul '09

Btw, i found that ShortIntegerArray missing the methods which providing a coercions to/from C. I made own subclass of it and implemented them there. It you want , i can send you the code. 2009/7/29 Igor Stasenko <siguctua(a)gmail.com>: > 2009/7/29 Igor Stasenko <siguctua(a)gmail.com>: >> 2009/7/29 David T. Lewis <lewis(a)mail.msen.com>: >>> >>> On Tue, Jul 28, 2009 at 06:39:50PM +0300, Igor Stasenko wrote: >>>> >>>> I'm using a smart syntax code gen for my plugin, >>>> and custom classes for auto-convert the oop<->c types >>>> >>>> and found that method: >>>> >>>> #ccg:emitLoadFor:from:on: >>>> >>>> is never get called by code generator (browser reports there is no >>>> senders of it). >>>> It is, however, implemented in multiple classes. >>>> >>>> As i understand, smart syntax plugin using: >>>> >>>> #ccg:prolog:expr:index: >>>> and >>>> #ccg:generateCoerceToValueFrom:on: >>>> >>>> for asking the classes how to load a value from stack or coerce from >>>> oop to C type. >>> >>> Interesting. This looks to me like some partially completed work in >>> the SmartSyntaxPluginCodeGenerator. The various implementations of >>> #ccg:emitLoadFor:from:on: call methods in the "linking" category >>> of SmartSyntaxPluginCodeGenerator, and a couple of these methods >>> are entirely unreferenced also. >>> >>> It looks as if the intent was to use #ccg:emitLoadFor:from:on: to >>> load things from the stack into local variables, but as you say it >>> is unreferenced, and all of these methods could presumably be >>> deleted without harm. >>> >>> The original implementation was by Andrew C Greenberg, and some >>> documentation is at wiki.squeak.org/squeak/850. >>> >>> Any suggestions as to whether this is work that should be completed >>> at some point (and therefore preserved for now), or if it should >>> just be deleted? >>> >> I found the following things useful: >> >> suppose i wrote a primitive: >> >> myPrimitive: argument oop >> > oops.. wrong method pattern :) > i meant: > > myPrimitive: argument arg2: oop > >> 1. self primitive: 'kaka' >> parameters: #(MyCustomType1 Oop). >> 2. x := oop asValue: MyCustomType2. >> 3. ^ (self cCode: 'foo(argument)') asOop: MyCustomType3 >> >> >> in (1), it generates the stack-load code by sending >> #ccg:prolog:expr:index: to MyCustomType class >> which implementation, in most cases, simply reusing the (2) >> >> in (2), its sending #ccg:generateCoerceToValueFrom:on: >> and your are to implement a coercion from Oop -> your custom value type >> >> in (3), it sends #ccg:generateCoerceToOopFrom:on: to get the code >> which converts the C-value to Oop. >> >> it is also sends #ccgDeclareCForVar: to retreive the argument >> declaration (this will be declared as a local variable in C primitive >> function). >> >> So, all cases seem covered perfectly: >> - argument declaration >> - Oop <-> C value coercion >> >> what else do we need? >> If i'm not mistaken - nothing more. I am done writing binding of a >> whole library API today, and not using anything except those :) >> >> >> -- >> Best regards, >> Igor Stasenko AKA sig. >> > > > > -- > Best regards, > Igor Stasenko AKA sig. > -- Best regards, Igor Stasenko AKA sig.

2 3

VMMaker: No senders of #ccg:emitLoadFor:from:on:
by Igor Stasenko 29 Jul '09

29 Jul '09

I'm using a smart syntax code gen for my plugin, and custom classes for auto-convert the oop<->c types and found that method: #ccg:emitLoadFor:from:on: is never get called by code generator (browser reports there is no senders of it). It is, however, implemented in multiple classes. As i understand, smart syntax plugin using: #ccg:prolog:expr:index: and #ccg:generateCoerceToValueFrom:on: for asking the classes how to load a value from stack or coerce from oop to C type. -- Best regards, Igor Stasenko AKA sig.

2 4

the initial memory allocation interface...
by Eliot Miranda 25 Jul '09

25 Jul '09

...stinks. I'm talking about sqAllocateMemory. Here's the base definition from platforms/Cross/vm/sq.h: #define sqAllocateMemory(minHeapSize, desiredHeapSize) malloc(desiredHeapSize) The problem here is that there's no obvious way for the client to know how much memory is returned. The signature implies it is somewhere between minHeapSize & desiredHeapSize inclusive (or null, if allocation failed). But how do you tell? Well, one could always ask to grow memory by 0 and see what you get back, except for the small chicken-and-egg problem that sqGrowMemory:By: and sqShrinkMemory:By: require the ammount of memory allocated as an argument: #define sqGrowMemoryBy(oldLimit, delta) oldLimit #define sqShrinkMemoryBy(oldLimit, delta) oldLimit So one has to go to extraordinary lengths that are completely non-obvious in the client code to actually pass-back the ammount allocated. Here's a client in readImageFromFile: "allocate a contiguous block of memory for the Squeak heap" memory *:=* self cCode: 'sqAllocateMemory(minimumMemory, heapSize)' . memory = nil ifTrue: [self insufficientMemoryAvailableError]. memStart *:=* self startOfMemory. self setMemoryLimit: (memStart + heapSize) - 24. "decrease memoryLimit a tad for safety" self setEndOfMemory: memStart + dataSize. and here's how one gets around the absence of a "how much was allocated" parameter, from platforms/Mac OS/vm/sqPlatformSpecifc.h #undef sqAllocateMemory usqInt sqAllocateMemoryMac(sqInt minHeapSize, sqInt *desiredHeapSize); #define sqAllocateMemory(x,y) sqAllocateMemoryMac(x,&y); Ingenious (seriously; John this is a very clever working around of the restriction). Totally non-obvious at the client site, but it still gets the job done without requiring the client to change. So this costs too much to maintain (a.k.a. puts too much strain on my rodent brain). We could go with an explicit out parameter, as in "allocate a contiguous block of memory for the Squeak heap" memory *:=* self cCode: 'sqAllocateMemory(minimumMemory, heapSize, &allocatedSize)'. memory = nil ifTrue: [self insufficientMemoryAvailableError]. memStart *:=* self startOfMemory. self setMemoryLimit: (memStart + allocatedSize) - 24. "decrease memoryLimit a tad for safety" self setEndOfMemory: memStart + dataSize. but out parameters are not supported by Smalltalk and the VM is a Smalltalk program goddammit. So I want to canvas opinions on the following simplified proposal, which is to drop the oldLimit parameter from sqGrowMemoryBy, discard sqShrinkMemoryBy (it is superfluous given sqGrowMemoryBy can take a signed argument) and use self sqGrowMemoryBy: 0 to determine how much memory has been allocated: "allocate a contiguous block of memory for the Squeak heap" memory *:=* self sqAllocateAtLeast: minimumMemory AtMostMemory: heapSize. memory = nil ifTrue: [self insufficientMemoryAvailableError]. heapSize *:=* self sqGrowMemoryBy: 0. memStart *:=* self startOfMemory. self setMemoryLimit: (memStart + heapSize) - 24. "decrease memoryLimit a tad for safety" self setEndOfMemory: memStart + dataSize. with the obvious default implementations in e.g. platforms/Cross/vm/sqMemory.c static unsigned long memoryLimit = 0; void * sqAllocateAtLeastAtMostMemory(unsigned long committed, unsigned long reserved) { void *mem; if ((mem = malloc(reserved)) ~= 0) memoryLimit = reserved; else if ((mem = malloc(committed)) ~= 0) memoryLimit = committed; return mem; } unsigned long sqGrowMemoryBy(sqInt delta) { return memoryLimit; } and everywhere in the VM which expects the old sqGrowMemoryBy to answer a pointer to the limit must instead add the result to the base of allocated memory; i.e. limit *:=* self sqGrowMemory: memoryLimit By: delta. limit = memoryLimit ifFalse: [self setMemoryLimit: limit - 24. "remove a tad for safety" self initializeMemoryFirstFree: freeBlock] is rewritten as limit *:=* memoryLimit + (self sqGrowMemoryBy: delta). limit = memoryLimit ifFalse: [self setMemoryLimit: limit - 24. "remove a tad for safety" self initializeMemoryFirstFree: freeBlock] Opinions? best Eliot

5 8

Fwd: [Pharo-project] Show stopper network bug on Windows
by Igor Stasenko 22 Jul '09

22 Jul '09

Hello guys, those, who not subscribed to the Pharo list. I think you should read this: http://code.google.com/p/pharo/issues/detail?id=961 since it seems there are some networking problems with Win32 VM. ---------- Forwarded message ---------- From: Igor Stasenko <siguctua(a)gmail.com> Date: 2009/7/22 Subject: Re: [Pharo-project] Show stopper network bug on Windows To: Pharo-project(a)lists.gforge.inria.fr 2009/7/22 Schwab,Wilhelm K <bschwab(a)anest.ufl.edu>: > > The lack of fully qualified names could easily be a windows stupidity. My Dolphin code for lookups has long been a little complicated: look up this, clear that, do it again, and it always seemed that I was fighting lack of imagination from the general direction of Redmond WA - who would have thought that possible? :) My favorite was win2k from (IIRC) pre-sp3: the only machine one could not find by name was the local host! > Concerning fully qualified host name. Yes. This is a well-known issue of windoze. For instance, my machine returns 'comp' on #localHostName request. Which is nothing more, than a computer name, i assigned in windoze settings. I didn't assigned my PC to be a part of any (sub)domain, so this is all it can report. The problems with dns/windoze host naming is comes back to Windows 3.11 era when there was no 'internet' but a local networks and windows used own identification mechanisms and protocols (NetBIOS, NetBEUI) to connect different hosts into a 'workgroup' :) However, next is getting interesting. I having a following network interfaces on my PC (i have to translate the text, because it prints things in russian) : (( The engineers at Redmond renamed the well-known unix ifconfig utility to ipconfig :) )) C:\>ipconfig /all Windows IP protocol settings Host name . . . . . . . . . : comp Primary DNS-suffix . . . . . . : Host type . . . . . . . . . . . . . : unknown IP-forwarding enabled . . . . : no WINS-proxy enabled . . . . . . . : no VMware Network Adapter VMnet8 - Ethernet adapter: DNS-suffix . . : IP-address . . . . . . . . . . . . : 192.168.6.1 Subnet mask . . . . . . . . . . : 255.255.255.0 Gateway . . . . . . . . . . : VMware Network Adapter VMnet1 - Ethernet adapter: DNS-suffix . . : IP-address . . . . . . . . . . . . : 192.168.89.1 Subnet mask . . . . . . . . . . : 255.255.255.0 Gateway . . . . . . . . . . : VirtualBox Host-Only Network - Ethernet adapter: DNS-suffix . . : IP-address . . . . . . . . . . . . : 192.168.56.1 Subnet mask . . . . . . . . . . : 255.255.255.0 Gateway . . . . . . . . . . : Network bridge - Ethernet adapter: DNS-suffix . . : IP-address . . . . . . . . . . . . : 192.168.0.3 Subnet mask . . . . . . . . . . : 255.255.255.0 Gateway . . . . . . . . . . : 192.168.0.1 Now, if i run: C:\>ping comp it starts pinging 192.168.56.1 address, which its thinks my extenal/local adapter address, but its obviously not i would expect that 'comp' should be resolved to 127.0.0.1 address (a loopback adapter) but windows thinks differently :) So, imagine i want to create a socket , listening for incoming connections at 'localHostName' address: NetNameResolver addressForName: NetNameResolver localHostName it gives me the same result as ping does: 192.168.56.1(192.168.56.1),0(0) surely, using this address i will never receive any incoming connections.. :) But what i didn't expected, it also opens a debug console and prints some kind of error about getnameinfo() function (again, translated, because my windoze is localized): getnameinfo: requested name is allowed and its found in database, but for this name the associated data is missing, which was allowed for it.. But its not producing a primitive failure. btw, same error shown in concole even if i do: NetNameResolver addressForName: 'localhost' i found that the error producing when i printing the instance of SocketAddress (#hostName method) weird... Next, nslookup utility of course, unable to resolve 'comp' name: C:\>nslookup Default Server: ns.ip.net.ua Address: 82.193.96.6 > comp Server: ns.ip.net.ua Address: 82.193.96.6 *** ns.ip.net.ua can't find comp: Non-existent domain > -- Best regards, Igor Stasenko AKA sig.

2 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev July 2009