I am working on a parser generator tool (a replacement for SmaCC) and
one of the things I a m interested in is the direct translation of
language specifications into the virtual machine code (SmaCC and my
current version of it use Squeak as the target language).

One of the problems I have is that, for some languages, the natural translation
into VM code uses computed gotos.
There are two scenarios here:

     1) goto X  where X is a variable.
     2) goto  (coll at: y)  where coll is a Collection.

For example, one such language is that of regular expressions, which I wish to translate into finite state machines implemented in VM code.
In this case I need case 2) gotos where coll is a collection of associations, possibly a
Dictionary. I also plan to write a debugger for this (and other languages) but that is another story.

I realize that the Cog VM is being built for Smalltalk (Squeak? Pharo?) for which the goto instructions are not needed and thus I assume unavailable. But there is something to
viewing a virtual machine as general purpose and thus the target of multiple languages as is
the case for the Java virtual machine.
If the Cog VM is viewed this way then I argue there is a need for my goto instructions
because some languages have need for them.
For example, many languages have case statements.  (I am all for object oriented
but I would be willing to accept a case statement in Smalltalk too;  the Squeak code
implemented one in Squeak doesn't cut it).

Anyway, I am not arguing to Change Squeak or Smalltalk but I am arguing
to have my goto instructions in Cog VM. Is there any chance of this?????

I don't know the Squeak VM or the Cog VM either but I assume these
instructions don't exist because I see no need of them when the source language is
Squeak or any version of Smalltalk for that matter. I also assume that there is already
a full list of 256 instructions in the Cog VM and thus no room for my goto instructions
unless some instructions are removed.

Are there Cog VM instructions that are so rarely used that they could be removed without
unreasonably slowing down the Cog VM interpretation of byte codes generated from Squeak source code?????

I accept that it will always be that almost all byte codes to be interpreted by the
Cog VM are generated from Smalltalk source.

Good luck with the Cog VM.  I look forward to seeing it used in Squeak/Pharo.

Ralph Boland

 

>>
>> I am working on a parser generator tool (a replacement for SmaCC) and
>> one of the things I a m interested in is the direct translation of
>> language specifications into the virtual machine code (SmaCC and my
>> current version of it use Squeak as the target language).
>>

> First, a different approach than compiling to Smalltalk is to compile to a
> parse tree.  We do this in the pseudo JavaScript compiler we've implemented
> at Cadence, translating an AST into a Squeak compiler parse tree for code
> generation.  Targeting a parse tree gives you much more freedom; you can
> express things that aren't expressible in Smalltalk.  And if you target
> bytecodes you can do even more.

I never considered using a parse tree as the target.  An interesting idea

which in many instances may be the best approach.  But for my regular expression

example I would still want to generate byte codes.  In any case I wouldn't want to
restrict users of my parser generator tool to any one of the three options (Smalltalk code,

parse tree, byte code).  It is my responsibility to make all three options as easy and
efficient as reasonably possible for users of the parser generator tool.  Haven't put
much thought into this yet though.   So far Smalltalk (Squeak) is the only option.

>> One of the problems I have is that, for some languages, the natural
>> translation
>> into VM code uses computed gotos.
>> There are two scenarios here:
>>
>>      1) goto X  where X is a variable.
>>      2) goto  (coll at: y)  where coll is a Collection.
>>

> There are several ways of implementing this without computed bytecodes in
> the instruction set, but there is also the possibility of implementing it
> directly in the instruction set.

> Off the top of my head one could

> - map to perform: using some mangled selector.  Yes this is problematic
> because one has to split the scope between the two methods, so in general
> it's not a solution

Doesn't appeal to me.

> - map to a case statement, which is what Squeak does. Just map it to a
> sequence of compare and branches.  Or better still, to a binary tree.
> Coincidentally this is used by the JIT to implement block dispatch in
> methods that contain more than one block.  I know of other VM
> implementations using it for PIC dispatch with really good performance.

I don't know what you mean my Squeak mapping to a case statement since
there is no case statement in  Squeak/Smalltalk and I can't off hand think of
where one is needed (some Squeak users might feel they need one but that is a

different matter). 

The use of compare and branches might be OK in some cases

but a mess for the finite state machines generated from regular expressions.

Actually, even with computed gotos  FSMs are somewhat messy but without them

it's worse.  I don't know what  'PIC dispatch' is.

To use a binary tree don't I need some kind of computed goto for when I reach
a leaf of the tree????

> - use thisContext pc: value.

This would be a possibility for me to experiment with for now.  When I have a working
parser generator tool I could campaign for my computed goto instructions to be added
to the VM. 

> This /should/ be fine in the stack VM, but
> slooooow in the JIT because internally mapping bytecode pcs to machine code
> pcs is slow, and currently slower still because the frame will be converted
> to a pure context and then converted back into a frame on the return from
> pc:.  But this solution isn't to be rejected out-of-hand.  It can be
> optimized to avoid the frame conversion and the JIT might be able to
> optimize it.

I assume that if computed gotos were used the translation to machine code would require a direct
mapping of (virtually labeled) bytecode locations to machine code locations.  I think this can be done
in a reasonable amount of time but others such as yourself clearly understand the issues far better than
I do.  The dirty solution to start would be to simply not JIT the code that uses computed gotos.

> The main problem is the compiler has no support for labels so
> there would be work here.

I don't mind doing the work but to my way of thinking  "goto X" is pretty basic
and is thus best handled at the VM/byte code level.  Anything else is doing
in a complicated way something that is fairly simple.  Of course changing

the VM/byte codes by even a single byte code is a major deal unless done
when the VM/byte codes are initially created.  Alas I must deal with what
already exists.  Even so, my preference is to work with the VM if at all

possible.

>> For example, one such language is that of regular expressions, which I
>> wish to translate into finite state machines implemented in VM code.
>> In this case I need case 2) gotos where coll is a collection of
>> associations, possibly a
>> Dictionary. I also plan to write a debugger for this (and other languages)
>> but that is another story.
>>
>> I realize that the Cog VM is being built for Smalltalk (Squeak? Pharo?)
>> for which the goto instructions are not needed and thus I assume
>> unavailable. But there is something to
>> viewing a virtual machine as general purpose and thus the target of
>> multiple languages as is
>> the case for the Java virtual machine.
>> If the Cog VM is viewed this way then I argue there is a need for my goto
>> instructions
>> because some languages have need for them.
>> For example, many languages have case statements.  (I am all for object
>> oriented
>> but I would be willing to accept a case statement in Smalltalk too;  the
>> Squeak code
>> implemented one in Squeak doesn't cut it).
>

> I've occasionally thought about this for many years.  A computed jump might
> be nice.  Eg index an Array literal of pcs with the integer on top of
> stack, falling through on bad type or out of range.

This is the way I am thinking.  If there are other reasons for a computed jumpTo
as well all the better.

>> Anyway, I am not arguing to Change Squeak or Smalltalk but I am arguing
>> to have my goto instructions in Cog VM. Is there any chance of this?????
>>

> There's no chance of me spending time implementing this any time soon.  I
> have too much high-priority tasks to tackle this.  But I want to encourage
> you or others to have a go implementing it.  It's fun!

I understand and am willing to be the one to add one or more computed jump
instructions, including working on the JIT code generator if needed.
As you say it should be fun (and also educational).  But

   1)  I am pretty busy now too and probably won't get to this for a year.

   2)  If I am to do this it would be great if someone can write a specification as to

        what is to be done.  If someone can write this now that would be great but

        if they write it when I post that I am ready to do the work that would also

        be fine.

   3)  I don't want to just have my own private VM/byte codes.  I want users of my

        parser generator tool to be able to load it into a standard version of Squeak

        and run it there including the possible generation of compilers for compiling
        their domain specific language programs into byte codes if desired.

>> I don't know the Squeak VM or the Cog VM either but I assume these
>> instructions don't exist because I see no need of them when the source
>> language is
>> Squeak or any version of Smalltalk for that matter. I also assume that
>> there is already
>> a full list of 256 instructions in the Cog VM and thus no room for my goto
>> instructions
>> unless some instructions are removed.
>>
>> Are there Cog VM instructions that are so rarely used that they could be
>> removed without
>> unreasonably slowing down the Cog VM interpretation of byte codes
>> generated from Squeak source code?????
>>

> The current set has 3 unused bytecodes, one of which Spur uses, so
> effectively there are two unused bytecodes.

Levente Uzonyi  in his posting pointed out that only one instruction is needed.

I don't like having to push the address to jump to onto the stack, preferring a byte
code with an argument, but I could live with his solution if that is what is decided.

In the case of  goto  coll at: X  the address is likely to end up on top of the stack
anyway so Levente's  jumpToTop instruction looks good in any case.

> The Cog VMs support multiple bytecode sets.  If you look at the
> BytecodeSets package on VMMaker you can read the class comments of the
> BytecodeEncoder subclasses such as EncoderForSistaV1.  These bytecode sets
> have a few more unused bytecodes.  This multiple bytecode set support is
> better implemented in Spur where there is only one compiled method header
> format and support for 64k literals.  So let me encourage you to move to
> Spur and to look at the Sista set.  The class comment of each encoder class
> specifies the instruction set it targets.

I am prepared to work with Spur and the Sista set.  I am looking for someone to
say that if I do this work that incorporating the work into Spur will be seriously
considered.

Ralph Boland

If the bytecode is one that takes an integer on top of stack, and an Array literal containing bytecode pcs, falling through on out of range, then nothing other than the index need be pushed on top of stack.  That would be my preference.

 

On Mon, Nov 3, 2014 at 11:34 AM, Eliot Miranda <[hidden email]> wrote:

 

If the bytecode is one that takes an integer on top of stack, and an Array literal containing bytecode pcs, falling through on out of range, then nothing other than the index need be pushed on top of stack.  That would be my preference.

I think Levante's suggestion for jumpToTop was just to take the pc from the top of the stack, with no indirection through an Array literal. Then jumping to a pc in an array would be 

- push the array

- push the index of the pc you want

- send #at:

- jumpToTop 

-Colin

 

> Hi Ralph,

>
> On Sun, Nov 2, 2014 at 9:48 PM, Ralph Boland <[hidden email]> wrote:
>
> >
> > >>
> > >> I am working on a parser generator tool (a replacement for SmaCC) and
> > >> one of the things I a m interested in is the direct translation of
> > >> language specifications into the virtual machine code (SmaCC and my
> > >> current version of it use Squeak as the target language).
> > >>
> >
> > > First, a different approach than compiling to Smalltalk is to compile to
> > a
> > > parse tree.  We do this in the pseudo JavaScript compiler we've
> > implemented
> > > at Cadence, translating an AST into a Squeak compiler parse tree for code
> > > generation.  Targeting a parse tree gives you much more freedom; you can
> > > express things that aren't expressible in Smalltalk.  And if you target
> > > bytecodes you can do even more.
> >
> >
> > I never considered using a parse tree as the target.  An interesting idea
> > which in many instances may be the best approach.  But for my regular
> > expression
> > example I would still want to generate byte codes.  In any case I wouldn't
> > want to
> > restrict users of my parser generator tool to any one of the three options
> > (Smalltalk code,
> > parse tree, byte code).  It is my responsibility to make all three options
> > as easy and
> > efficient as reasonably possible for users of the parser generator tool.
> > Haven't put
> > much thought into this yet though.   So far Smalltalk (Squeak) is the only
> > option.
> >
> >
> > >> One of the problems I have is that, for some languages, the natural
> > >> translation
> > >> into VM code uses computed gotos.
> > >> There are two scenarios here:
> > >>
> > >>      1) goto X  where X is a variable.
> > >>      2) goto  (coll at: y)  where coll is a Collection.
> > >>
> >
> > > There are several ways of implementing this without computed bytecodes in
> > > the instruction set, but there is also the possibility of implementing it
> > > directly in the instruction set.
> >
> > > Off the top of my head one could
> >
> > > - map to perform: using some mangled selector.  Yes this is problematic
> > > because one has to split the scope between the two methods, so in general
> > > it's not a solution
> >
> > Doesn't appeal to me.
> >
> > > - map to a case statement, which is what Squeak does. Just map it to a
> > > sequence of compare and branches.  Or better still, to a binary tree.
> > > Coincidentally this is used by the JIT to implement block dispatch in
> > > methods that contain more than one block.  I know of other VM
> > > implementations using it for PIC dispatch with really good performance.
> >
> > I don't know what you mean my Squeak mapping to a case statement since
> > there is no case statement in  Squeak/Smalltalk and I can't off hand think
> > of
> > where one is needed (some Squeak users might feel they need one but that
> > is a
> > different matter).
> >
>
> But there is.  And it is very convenient when one doesn't want to spread a
> case over different classes, or when the cases distribute over values of
> the same class (e.g. integer values).  People often claim that a case
> statement isn't necessary because one has polymorphism, but unless the
> language supports singletons (which of course Smalltalk does not) a case
> statement is much more readable than e.g. an if-then-else tree or a
> "Dictionary mapping values to blocks or selectors" solution.
>
> Since you're unaware of the case statement I suggest you browse senders of
> caseOf: and caseOf:otherwise: in Squeak trunk (which includes selectors of
> optimized selectors that end up not being sent) or caseError, which is sent
> when there's no otherwise clause.  Here's an example:
>
> menuHook: aMenu named: aSymbol shifted: aBool
> "Enhance aMenu with registered services."
> aSymbol
> caseOf:
> { [ #classListMenu ] -> [ ServiceGui browser: self classMenu: aMenu ].
> [ #codePaneMenu ] -> [ ServiceGui browser: self codePaneMenu: aMenu ].
> [ #messageCategoryMenu] -> [ ServiceGui browser: self messageCategoryMenu:
> aMenu ].
> [ #messageListMenu ] -> [ ServiceGui browser: self messageListMenu: aMenu ].
> [ #systemCategoryMenu ] -> [ ServiceGui browser: self classCategoryMenu:
> aMenu ] }
> otherwise: [ "do nothing" ]
>
> This compiles down to a sequence of comparisons.

I was aware of caseOf: in Squeak.  I always found it awkward to use and
felt a true case statement would be simpler.  Alas, it's impossible to
have a true case statement added to Smalltalk now I think.

caseOf: is user code and thus, in principle, outside the scope of the VM.
For example, if I didn't want the sequential search of caseOf:  I could
implement a method  doBlockAt:  method in class Dictionary that assumes
the values stored in the dictionary are blocks and invokes the block
found by performing at: on the arg of doBlockAt:.
But the code for both caseOf: and doBlockAt: are written by users so the
VM can't optimize it, is forced to have the overhead of using blocks,
and must be locked in to the search algorithm the user has chosen (for
better or for worse).

> > The use of compare and branches might be OK in some cases
> > but a mess for the finite state machines generated from regular
> > expressions.
> > Actually, even with computed gotos  FSMs are somewhat messy but without
> > them
> > it's worse.  I don't know what  'PIC dispatch' is.
> >
>
> Polymorphic inline cache dispatch.  In JIT VMs such as Cog polymorphic send
> sites are optimized using jump tables, one jump for each class encountered
> at the send site, up tio some small limit such as 6 cases.  Right now in
> Cog these jump tables are  sequential comparisons.  But in some VMs, where
> the degree of polymorphism may be much higher (think prototype languages
> such as JavaScript) a binary tree may be much miore efficient, if harder to
> organize.
>
> To use a binary tree don't I need some kind of computed goto for when I
> > reach
> > a leaf of the tree????
> >
>
> No.  Once you're at the leaf you just include the bytecodes.  So for
> example take the following case statement:
>
> quickPrimitiveGeneratorFor: aQuickPrimitiveIndex
> <api>
> <returnTypeC: 'int (*quickPrimitiveGeneratorFor(sqInt
> aQuickPrimitiveIndex))(void)'>
> ^aQuickPrimitiveIndex
> caseOf: {
> [256] -> [#genQuickReturnSelf].
> [257] -> [#genQuickReturnConstNil].
> [258] -> [#genQuickReturnConstTrue].
> [259] -> [#genQuickReturnConstFalse].
> [260] -> [#genQuickReturnConstMinusOne].
> [261] -> [#genQuickReturnConstZero].
> [262] -> [#genQuickReturnConstOne].
> [263] -> [#genQuickReturnConstTwo] }
> otherwise: [#genQuickReturnInstVar]
>
> This is compiled in the current Squeak compiler as
>
> 61 <10> pushTemp: 0
> 62 <88> dup
> 63 <2A> pushConstant: 256
> 64 <B6> send: =
> 65 <9B> jumpFalse: 70
> 66 <87> pop
> 67 <29> pushConstant: #genQuickReturnSelf
> 68 <A4 35> jumpTo: 123
> 70 <88> dup
> 71 <28> pushConstant: 257
> 72 <B6> send: =
> 73 <9B> jumpFalse: 78
> 74 <87> pop
> 75 <21> pushConstant: #genQuickReturnConstNil
> 76 <A4 2D> jumpTo: 123
> ...
> 117 <22> pushConstant: 263
> 118 <B6> send: =
> 119 <99> jumpFalse: 122
> 120 <21> pushConstant: #genQuickReturnConstTwo
> 121 <90> jumpTo: 123
> 122 <20> pushConstant: #genQuickReturnInstVar
> 123 <7C> returnTop
>
> but it could be more efficient if the code were
>
>     pushTemp: 0
>     dup
>     pushConstant: 256
>     send: <
>     jumpFalse: L1
>     dup
>     pushConstant: 260
>     send: <
>     jumpFalse: L2
>     dup
>     pushConstant: 258
>     send: <
>     jumpFalse: L3
>     dup
>     pushConstant: 257
>     send: <
>     jumpFalse: L4
>     pushConstant: #genQuickReturnSelf
>     jumpTo: L0
>     pushConstant: #genQuickReturnConstNil
>     jumpTo: L0
>     ...
>    L1:
>     pushConstant: #genQuickReturnInstVar
>    L0:
>     returnTop

I have no problem with this code per say; if Squeak, the language,
had a case statement it could be implemented this way.

But I wouldn't want to be forced to implement my FSMs this way.
It might be acceptable for small FSMs.
I want to avoid sequential search and
 even binary search might be rather expensive.
I look at computed gotos as the solution but,
as you pointed out, computed gotos pose problems for JIT.
Admittedly, for large FSM's, it might be best or necessary to
use a FSM simulator anyway, as I do now.

> > - use thisContext pc: value.
> >
> > This would be a possibility for me to experiment with for now.  When I
> > have a working
> > parser generator tool I could campaign for my computed goto instructions
> > to be added
> > to the VM.
> >
> > > This /should/ be fine in the stack VM, but
> > > slooooow in the JIT because internally mapping bytecode pcs to machine
> > code
> > > pcs is slow, and currently slower still because the frame will be
> > converted
> > > to a pure context and then converted back into a frame on the return from
> > > pc:.  But this solution isn't to be rejected out-of-hand.  It can be
> > > optimized to avoid the frame conversion and the JIT might be able to
> > > optimize it.
> >
> > I assume that if computed gotos were used the translation to machine code
> > would require a direct
> > mapping of (virtually labeled) bytecode locations to machine code
> > locations.  I think this can be done
> > in a reasonable amount of time but others such as yourself clearly
> > understand the issues far better than
> > I do.  The dirty solution to start would be to simply not JIT the code
> > that uses computed gotos.
> >
>
> Yes, it could be.  The JIT would generate a jump table from the literal
> containing the bytecoded pcs, and there would be no conversion; only
> indexing the jump table and jumping.  Again, organizing that table as a
> binary switch may well be faster on modern architectures.  Indirect jumps
> typically involve pipeline stalls, whereas binary jump trees don't.
>
> > The main problem is the compiler has no support for labels so
> > > there would be work here.
> >
> > I don't mind doing the work but to my way of thinking  "goto X" is pretty
> > basic
> > and is thus best handled at the VM/byte code level.  Anything else is doing
> > in a complicated way something that is fairly simple.  Of course changing
> > the VM/byte codes by even a single byte code is a major deal unless done
> > when the VM/byte codes are initially created.  Alas I must deal with what
> > already exists.  Even so, my preference is to work with the VM if at all
> > possible.
> >
> >
> > >> For example, one such language is that of regular expressions, which I
> > >> wish to translate into finite state machines implemented in VM code.
> > >> In this case I need case 2) gotos where coll is a collection of
> > >> associations, possibly a
> > >> Dictionary. I also plan to write a debugger for this (and other
> > languages)
> > >> but that is another story.
> > >>
> > >> I realize that the Cog VM is being built for Smalltalk (Squeak? Pharo?)
> > >> for which the goto instructions are not needed and thus I assume
> > >> unavailable. But there is something to
> > >> viewing a virtual machine as general purpose and thus the target of
> > >> multiple languages as is
> > >> the case for the Java virtual machine.
> > >> If the Cog VM is viewed this way then I argue there is a need for my
> > goto
> > >> instructions
> > >> because some languages have need for them.
> > >> For example, many languages have case statements.  (I am all for object
> > >> oriented
> > >> but I would be willing to accept a case statement in Smalltalk too;  the
> > >> Squeak code
> > >> implemented one in Squeak doesn't cut it).
> > >
> >
> > > I've occasionally thought about this for many years.  A computed jump
> > might
> > > be nice.  Eg index an Array literal of pcs with the integer on top of
> > > stack, falling through on bad type or out of range.
> >
> > This is the way I am thinking.  If there are other reasons for a computed
> > jumpTo
> > as well all the better.
> >
> > >> Anyway, I am not arguing to Change Squeak or Smalltalk but I am arguing
> > >> to have my goto instructions in Cog VM. Is there any chance of this?????
> > >>
> >
> > > There's no chance of me spending time implementing this any time soon.  I
> > > have too much high-priority tasks to tackle this.  But I want to
> > encourage
> > > you or others to have a go implementing it.  It's fun!
> >
> > I understand and am willing to be the one to add one or more computed jump
> > instructions, including working on the JIT code generator if needed.
> > As you say it should be fun (and also educational).  But
> >    1)  I am pretty busy now too and probably won't get to this for a year.
> >    2)  If I am to do this it would be great if someone can write a
> > specification as to
> >         what is to be done.  If someone can write this now that would be
> > great but
> >         if they write it when I post that I am ready to do the work that
> > would also
> >         be fine.
> >    3)  I don't want to just have my own private VM/byte codes.  I want
> > users of my
> >         parser generator tool to be able to load it into a standard
> > version of Squeak
> >         and run it there including the possible generation of compilers
> > for compiling
> >         their domain specific language programs into byte codes if desired.
> >
>
> Sure.  Get back to me when you're ready to work on it and I'll write the
> specification, and help you with whatever info you need.  There should be
> room in the bytecode sets we'll likely be using next year.

WILL DO.  SHOULD BE FUN.
WILL DO.  SHOULD BE FUN.
WILL DO.  SHOULD BE FUN.

>

> >
> > >> I don't know the Squeak VM or the Cog VM either but I assume these
> > >> instructions don't exist because I see no need of them when the source
> > >> language is
> > >> Squeak or any version of Smalltalk for that matter. I also assume that
> > >> there is already
> > >> a full list of 256 instructions in the Cog VM and thus no room for my
> > goto
> > >> instructions
> > >> unless some instructions are removed.
> > >>
> > >> Are there Cog VM instructions that are so rarely used that they could be
> > >> removed without
> > >> unreasonably slowing down the Cog VM interpretation of byte codes
> > >> generated from Squeak source code?????
> > >>
> >
> > > The current set has 3 unused bytecodes, one of which Spur uses, so
> > > effectively there are two unused bytecodes.
> >
> > Levente Uzonyi  in his posting pointed out that only one instruction is
> > needed.
> > I don't like having to push the address to jump to onto the stack,
> > preferring a byte
> > code with an argument, but I could live with his solution if that is what
> > is decided.
> > In the case of  goto  coll at: X  the address is likely to end up on top
> > of the stack
> > anyway so Levente's  jumpToTop instruction looks good in any case.
> >
>
> If the bytecode is one that takes an integer on top of stack, and an Array
> literal containing bytecode pcs, falling through on out of range, then
> nothing other than the index need be pushed on top of stack.  That would be
> my preference.
>

Again, for my FSM, case this would often be considered to be good.
But if the state transition tables are sparse then Dictionaries
might be preferable to Arrays.

My expection is that  at:  be sent to the collection object
 to get the address to go to.  Knowing that the collection
is an array though makes it easier for the compiler/VM to
ensure that the addresses stored in the collection are valid.
Actually, the compiler will be generating the addresses.
Does the VM have absolute trust in the compiler to generate valid addresses?

What if the compiler is generated by my parser generator tool?

> > The Cog VMs support multiple bytecode sets.  If you look at the
> > > BytecodeSets package on VMMaker you can read the class comments of the
> > > BytecodeEncoder subclasses such as EncoderForSistaV1.  These bytecode
> > sets
> > > have a few more unused bytecodes.  This multiple bytecode set support is
> > > better implemented in Spur where there is only one compiled method header
> > > format and support for 64k literals.  So let me encourage you to move to
> > > Spur and to look at the Sista set.  The class comment of each encoder
> > class
> > > specifies the instruction set it targets.
> >
> > I am prepared to work with Spur and the Sista set.  I am looking for
> > someone to
> > say that if I do this work that incorporating the work into Spur will be
> > seriously
> > considered.
> >
>
> I can say that, yes.  I will seriously consider adding it.  I think its a
> useful thing in the bytecode set, precisely to allow compiling other
> languages to the VM.

GLAD TO HEAR THIS.

>
>
>
> >
> > Ralph Boland
> >
> >
> >
>
>
> --
> best,
> Eliot

Ralph

On 3 November 2014 12:34, <[hidden email]> wrote:

Send Vm-dev mailing list submissions to
        [hidden email]

To subscribe or unsubscribe via the World Wide Web, visit
        http://lists.squeakfoundation.org/mailman/listinfo/vm-dev
or, via email, send a message with subject or body 'help' to
        [hidden email]

You can reach the person managing the list at
        [hidden email]

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Vm-dev digest..."


Today's Topics:

   1. Re: goto instruction with Cog VM (Eliot Miranda)


----------------------------------------------------------------------

Message: 1
Date: Mon, 3 Nov 2014 11:34:48 -0800
From: Eliot Miranda <[hidden email]>
Subject: Re: [Vm-dev] goto instruction with Cog VM
To: Squeak Virtual Machine Development Discussion
        <[hidden email]>
Message-ID:
        <[hidden email]>
Content-Type: text/plain; charset="utf-8"

Hi Ralph,

On Sun, Nov 2, 2014 at 9:48 PM, Ralph Boland <[hidden email]> wrote:

>
> >>
> >> I am working on a parser generator tool (a replacement for SmaCC) and
> >> one of the things I a m interested in is the direct translation of
> >> language specifications into the virtual machine code (SmaCC and my
> >> current version of it use Squeak as the target language).
> >>
>
> > First, a different approach than compiling to Smalltalk is to compile to
> a
> > parse tree.  We do this in the pseudo JavaScript compiler we've
> implemented
> > at Cadence, translating an AST into a Squeak compiler parse tree for code
> > generation.  Targeting a parse tree gives you much more freedom; you can
> > express things that aren't expressible in Smalltalk.  And if you target
> > bytecodes you can do even more.
>
>
> I never considered using a parse tree as the target.  An interesting idea
> which in many instances may be the best approach.  But for my regular
> expression
> example I would still want to generate byte codes.  In any case I wouldn't
> want to
> restrict users of my parser generator tool to any one of the three options
> (Smalltalk code,
> parse tree, byte code).  It is my responsibility to make all three options
> as easy and
> efficient as reasonably possible for users of the parser generator tool.
> Haven't put
> much thought into this yet though.   So far Smalltalk (Squeak) is the only
> option.
>
>
> >> One of the problems I have is that, for some languages, the natural
> >> translation
> >> into VM code uses computed gotos.
> >> There are two scenarios here:
> >>
> >>      1) goto X  where X is a variable.
> >>      2) goto  (coll at: y)  where coll is a Collection.
> >>
>
> > There are several ways of implementing this without computed bytecodes in
> > the instruction set, but there is also the possibility of implementing it
> > directly in the instruction set.
>
> > Off the top of my head one could
>
> > - map to perform: using some mangled selector.  Yes this is problematic
> > because one has to split the scope between the two methods, so in general
> > it's not a solution
>
> Doesn't appeal to me.
>
> > - map to a case statement, which is what Squeak does. Just map it to a
> > sequence of compare and branches.  Or better still, to a binary tree.
> > Coincidentally this is used by the JIT to implement block dispatch in
> > methods that contain more than one block.  I know of other VM
> > implementations using it for PIC dispatch with really good performance.
>
> I don't know what you mean my Squeak mapping to a case statement since
> there is no case statement in  Squeak/Smalltalk and I can't off hand think
> of
> where one is needed (some Squeak users might feel they need one but that
> is a
> different matter).
>

But there is.  And it is very convenient when one doesn't want to spread a
case over different classes, or when the cases distribute over values of
the same class (e.g. integer values).  People often claim that a case
statement isn't necessary because one has polymorphism, but unless the
language supports singletons (which of course Smalltalk does not) a case
statement is much more readable than e.g. an if-then-else tree or a
"Dictionary mapping values to blocks or selectors" solution.

Since you're unaware of the case statement I suggest you browse senders of
caseOf: and caseOf:otherwise: in Squeak trunk (which includes selectors of
optimized selectors that end up not being sent) or caseError, which is sent
when there's no otherwise clause.  Here's an example:

menuHook: aMenu named: aSymbol shifted: aBool
"Enhance aMenu with registered services."
aSymbol
caseOf:
{ [ #classListMenu ] -> [ ServiceGui browser: self classMenu: aMenu ].
[ #codePaneMenu ] -> [ ServiceGui browser: self codePaneMenu: aMenu ].
[ #messageCategoryMenu] -> [ ServiceGui browser: self messageCategoryMenu:
aMenu ].
[ #messageListMenu ] -> [ ServiceGui browser: self messageListMenu: aMenu ].
[ #systemCategoryMenu ] -> [ ServiceGui browser: self classCategoryMenu:
aMenu ] }
otherwise: [ "do nothing" ]

This compiles down to a sequence of comparisons.

The use of compare and branches might be OK in some cases
> but a mess for the finite state machines generated from regular
> expressions.
> Actually, even with computed gotos  FSMs are somewhat messy but without
> them
> it's worse.  I don't know what  'PIC dispatch' is.
>

Polymorphic inline cache dispatch.  In JIT VMs such as Cog polymorphic send
sites are optimized using jump tables, one jump for each class encountered
at the send site, up tio some small limit such as 6 cases.  Right now in
Cog these jump tables are  sequential comparisons.  But in some VMs, where
the degree of polymorphism may be much higher (think prototype languages
such as JavaScript) a binary tree may be much miore efficient, if harder to
organize.

To use a binary tree don't I need some kind of computed goto for when I
> reach
> a leaf of the tree????
>

No.  Once you're at the leaf you just include the bytecodes.  So for
example take the following case statement:

quickPrimitiveGeneratorFor: aQuickPrimitiveIndex
<api>
<returnTypeC: 'int (*quickPrimitiveGeneratorFor(sqInt
aQuickPrimitiveIndex))(void)'>
^aQuickPrimitiveIndex
caseOf: {
[256] -> [#genQuickReturnSelf].
[257] -> [#genQuickReturnConstNil].
[258] -> [#genQuickReturnConstTrue].
[259] -> [#genQuickReturnConstFalse].
[260] -> [#genQuickReturnConstMinusOne].
[261] -> [#genQuickReturnConstZero].
[262] -> [#genQuickReturnConstOne].
[263] -> [#genQuickReturnConstTwo] }
otherwise: [#genQuickReturnInstVar]

This is compiled in the current Squeak compiler as

61 <10> pushTemp: 0
62 <88> dup
63 <2A> pushConstant: 256
64 <B6> send: =
65 <9B> jumpFalse: 70
66 <87> pop
67 <29> pushConstant: #genQuickReturnSelf
68 <A4 35> jumpTo: 123
70 <88> dup
71 <28> pushConstant: 257
72 <B6> send: =
73 <9B> jumpFalse: 78
74 <87> pop
75 <21> pushConstant: #genQuickReturnConstNil
76 <A4 2D> jumpTo: 123
...
117 <22> pushConstant: 263
118 <B6> send: =
119 <99> jumpFalse: 122
120 <21> pushConstant: #genQuickReturnConstTwo
121 <90> jumpTo: 123
122 <20> pushConstant: #genQuickReturnInstVar
123 <7C> returnTop

but it could be more efficient if the code were

    pushTemp: 0
    dup
    pushConstant: 256
    send: <
    jumpFalse: L1
    dup
    pushConstant: 260
    send: <
    jumpFalse: L2
    dup
    pushConstant: 258
    send: <
    jumpFalse: L3
    dup
    pushConstant: 257
    send: <
    jumpFalse: L4
    pushConstant: #genQuickReturnSelf
    jumpTo: L0
    pushConstant: #genQuickReturnConstNil
    jumpTo: L0
    ...
   L1:
    pushConstant: #genQuickReturnInstVar
   L0:
    returnTop

> - use thisContext pc: value.
>
> This would be a possibility for me to experiment with for now.  When I
> have a working
> parser generator tool I could campaign for my computed goto instructions
> to be added
> to the VM.
>
> > This /should/ be fine in the stack VM, but
> > slooooow in the JIT because internally mapping bytecode pcs to machine
> code
> > pcs is slow, and currently slower still because the frame will be
> converted
> > to a pure context and then converted back into a frame on the return from
> > pc:.  But this solution isn't to be rejected out-of-hand.  It can be
> > optimized to avoid the frame conversion and the JIT might be able to
> > optimize it.
>
> I assume that if computed gotos were used the translation to machine code
> would require a direct
> mapping of (virtually labeled) bytecode locations to machine code
> locations.  I think this can be done
> in a reasonable amount of time but others such as yourself clearly
> understand the issues far better than
> I do.  The dirty solution to start would be to simply not JIT the code
> that uses computed gotos.
>

Yes, it could be.  The JIT would generate a jump table from the literal
containing the bytecoded pcs, and there would be no conversion; only
indexing the jump table and jumping.  Again, organizing that table as a
binary switch may well be faster on modern architectures.  Indirect jumps
typically involve pipeline stalls, whereas binary jump trees don't.

> The main problem is the compiler has no support for labels so
> > there would be work here.
>
> I don't mind doing the work but to my way of thinking  "goto X" is pretty
> basic
> and is thus best handled at the VM/byte code level.  Anything else is doing
> in a complicated way something that is fairly simple.  Of course changing
> the VM/byte codes by even a single byte code is a major deal unless done
> when the VM/byte codes are initially created.  Alas I must deal with what
> already exists.  Even so, my preference is to work with the VM if at all
> possible.
>
>
> >> For example, one such language is that of regular expressions, which I
> >> wish to translate into finite state machines implemented in VM code.
> >> In this case I need case 2) gotos where coll is a collection of
> >> associations, possibly a
> >> Dictionary. I also plan to write a debugger for this (and other
> languages)
> >> but that is another story.
> >>
> >> I realize that the Cog VM is being built for Smalltalk (Squeak? Pharo?)
> >> for which the goto instructions are not needed and thus I assume
> >> unavailable. But there is something to
> >> viewing a virtual machine as general purpose and thus the target of
> >> multiple languages as is
> >> the case for the Java virtual machine.
> >> If the Cog VM is viewed this way then I argue there is a need for my
> goto
> >> instructions
> >> because some languages have need for them.
> >> For example, many languages have case statements.  (I am all for object
> >> oriented
> >> but I would be willing to accept a case statement in Smalltalk too;  the
> >> Squeak code
> >> implemented one in Squeak doesn't cut it).
> >
>
> > I've occasionally thought about this for many years.  A computed jump
> might
> > be nice.  Eg index an Array literal of pcs with the integer on top of
> > stack, falling through on bad type or out of range.
>
> This is the way I am thinking.  If there are other reasons for a computed
> jumpTo
> as well all the better.
>
> >> Anyway, I am not arguing to Change Squeak or Smalltalk but I am arguing
> >> to have my goto instructions in Cog VM. Is there any chance of this?????
> >>
>
> > There's no chance of me spending time implementing this any time soon.  I
> > have too much high-priority tasks to tackle this.  But I want to
> encourage
> > you or others to have a go implementing it.  It's fun!
>
> I understand and am willing to be the one to add one or more computed jump
> instructions, including working on the JIT code generator if needed.
> As you say it should be fun (and also educational).  But
>    1)  I am pretty busy now too and probably won't get to this for a year.
>    2)  If I am to do this it would be great if someone can write a
> specification as to
>         what is to be done.  If someone can write this now that would be
> great but
>         if they write it when I post that I am ready to do the work that
> would also
>         be fine.
>    3)  I don't want to just have my own private VM/byte codes.  I want
> users of my
>         parser generator tool to be able to load it into a standard
> version of Squeak
>         and run it there including the possible generation of compilers
> for compiling
>         their domain specific language programs into byte codes if desired.
>

Sure.  Get back to me when you're ready to work on it and I'll write the
specification, and help you with whatever info you need.  There should be
room in the bytecode sets we'll likely be using next year.


>
> >> I don't know the Squeak VM or the Cog VM either but I assume these
> >> instructions don't exist because I see no need of them when the source
> >> language is
> >> Squeak or any version of Smalltalk for that matter. I also assume that
> >> there is already
> >> a full list of 256 instructions in the Cog VM and thus no room for my
> goto
> >> instructions
> >> unless some instructions are removed.
> >>
> >> Are there Cog VM instructions that are so rarely used that they could be
> >> removed without
> >> unreasonably slowing down the Cog VM interpretation of byte codes
> >> generated from Squeak source code?????
> >>
>
> > The current set has 3 unused bytecodes, one of which Spur uses, so
> > effectively there are two unused bytecodes.
>
> Levente Uzonyi  in his posting pointed out that only one instruction is
> needed.
> I don't like having to push the address to jump to onto the stack,
> preferring a byte
> code with an argument, but I could live with his solution if that is what
> is decided.
> In the case of  goto  coll at: X  the address is likely to end up on top
> of the stack
> anyway so Levente's  jumpToTop instruction looks good in any case.
>

If the bytecode is one that takes an integer on top of stack, and an Array
literal containing bytecode pcs, falling through on out of range, then
nothing other than the index need be pushed on top of stack.  That would be
my preference.

> The Cog VMs support multiple bytecode sets.  If you look at the
> > BytecodeSets package on VMMaker you can read the class comments of the
> > BytecodeEncoder subclasses such as EncoderForSistaV1.  These bytecode
> sets
> > have a few more unused bytecodes.  This multiple bytecode set support is
> > better implemented in Spur where there is only one compiled method header
> > format and support for 64k literals.  So let me encourage you to move to
> > Spur and to look at the Sista set.  The class comment of each encoder
> class
> > specifies the instruction set it targets.
>
> I am prepared to work with Spur and the Sista set.  I am looking for
> someone to
> say that if I do this work that incorporating the work into Spur will be
> seriously
> considered.
>

I can say that, yes.  I will seriously consider adding it.  I think its a
useful thing in the bytecode set, precisely to allow compiling other
languages to the VM.



>
> Ralph Boland
>
>
>


--
best,
Eliot
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20141103/469369c9/attachment.htm

------------------------------

_______________________________________________
Vm-dev mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/vm-dev


End of Vm-dev Digest, Vol 101, Issue 7
**************************************

--

-- Those who can make you believe absurdities, can make you commit atrocities -- Voltaire

 

> Hi Ralph,
...

> >
> > I was aware of caseOf: in Squeak.  I always found it awkward to use and
> > felt a true case statement would be simpler.  Alas, it's impossible to
> > have a true case statement added to Smalltalk now I think.

> So what's a "true" case statement?  For me, at least, the Squeak one *is*,
> and is more general than one limited to purely integer keys, as for example
> is C's switch statement.  A number of languages provide case statements
> that are like Squeak's.  What do you consider a "true" case statement?

I mean that:  caseOf: is not part of the language itself but rather part of the
standard library or set of packages that one finds in the IDE.  To be part of the
language it would need to be something the compiler is aware of. 

That is to
day the Smalltalk language is not very much.  Smalltalk (Squeak) the language
would not include Sets or Dictionaries but would include (some) Array classes
because some aspects of Arrays are dealt with directly by the compiler.

Selectors such as  ifTrue: and  to:do:  are part of the language because they are inlined by the compiler.

Put another way,  if I could get my doBlockAt: method incorporated into the Squeak IDE

it would nevertheless NOT be part of Squeak the language.

The consequence of  caseOf:  not being part of the language is that the compiler/VM

cannot perform optimizations when caseOf:  is run into but must treat it as

user written code.

Squeak's  caseOf:  is more general than C's switch statement but it could be more
general in that there is a hard coded message (=).  I would like to be able to replace

the '=' message by an arbitrary binary operator such as  includes:  or '>'.

I have to backtrack here:  I looked at the code and it looks like the compiler inlines

caseOf:  and caseOf:otherwise.  If so then these selectors are part of the language

by my definition.

...

> > But I wouldn't want to be forced to implement my FSMs this way.
> > It might be acceptable for small FSMs.
> > I want to avoid sequential search and
> >  even binary search might be rather expensive.
> > I look at computed gotos as the solution but,
> > as you pointed out, computed gotos pose problems for JIT.
> > Admittedly, for large FSM's, it might be best or necessary to
> > use a FSM simulator anyway, as I do now.


> Nah.  One should always be able to map it down somehow.  Tis will be easier
> with the Spur instruction set which lifts number of literals and length of
> branches limits.

Good to hear.


> > Again, for my FSM, case this would often be considered to be good.
> > But if the state transition tables are sparse then Dictionaries
> > might be preferable to Arrays.
>

> Yes, but getting to the limit of what the VM can reasonably interpret.
> Better would be an Array of value. pc pairs, where the keys are the values
> the switch bytecode compares top of stack against, and the pcs are where to
> jump to on a match.  The JIT can therefore implement the table as it sees
> fit, whereas the interpreter can just do a linear search through the Array.

I am looking at this from the point of view of a compiler writer/generator and consider

your proposal as inadequate for my needs.  You, I think, are looking at this from
the point of view of a VM writer and what can reasonably be delivered.  I don't think

what I want is overly difficult for the interpreter to deliver but as you pointed out,

and you know much better than I, what I want causes serious problems for the VM.

> > My expection is that  at:  be sent to the collection object
> >  to get the address to go to.  Knowing that the collection
> > is an array though makes it easier for the compiler/VM to
> > ensure that the addresses stored in the collection are valid.
> > Actually, the compiler will be generating the addresses.
> > Does the VM have absolute trust in the compiler to generate valid
> > addresses?


> Yes.  Generate bad bytecode and the VM crashes.
 

This is what I expected to hear but wanted it to be clear for compilers generated
by my parser generator tool as you did.

Ralph