September 2010 - squirrelfish-dev - lists.macosforge.org

Ever considered inlining 'hot' functions?
by Holger Freyther 21 Sep '10

21 Sep '10

Hi all, I might be completely wrong, my understanding of our JIT or JIT in general might be flawed as well. I have been reading some papers on SELF93s type based optimizations and related papers and started to wonder. >From my understanding our JIT is generating code on the method level (a whole method), did you ever consider inlining small functions? E.g. where the number of bytecode is < 2 * the size of an activation record? is that already done? was it tried? does it make sense? z.

3 6

Re: [squirrelfish] Generating bytecode documentation from Interpreter.cpp
by Maciej Stachowiak 21 Sep '10

21 Sep '10

On Sep 21, 2010, at 11:22 AM, Holger Freyther wrote: > On 09/22/2010 01:38 AM, Maciej Stachowiak wrote: > >>> comments? >> >> The docs actually are generated from Interpreter.cpp, from the doc comments on each case statement. It just hasn't been updated in a while. Also the following opcodes lack documentation: > > Could you help me to find that script? I can help with describing the semantic > of these opcodes. This rule from DerivedSources.make is supposed to regenerate it: docs/bytecode.html: make-bytecode-docs.pl Interpreter.cpp perl $^ $@ The script is at docs/make-butecode-docs.pl Regards, Maciej

1 0

Re: [squirrelfish] Ever considered inlining 'hot' functions?
by Oliver Hunt 21 Sep '10

21 Sep '10

arguments is a live reference to the arguments on the call stack, so function f(a) { g(); alert([a, arguments[0]]) } function g() { f.arguments[0] = "foo"; } f("bar"); will produce foo, foo Isn't ES a marvelous language? :D caller and callee aren't writable though. --Oliver On Sep 21, 2010, at 11:37 AM, Holger Freyther wrote: > On 09/22/2010 02:09 AM, Oliver Hunt wrote: > >> >> In the general case this would require being able to convert from inlined code to a 'correct' call stack at some arbitrary point > > argh, I had hoped that such things would be only necessary when we have a > debugger attached. This mean besides what maciej had said, we will need to > remember that this range of instructions belongs to a inlined call and be able > to generate the virtual call record. > > Can one write to arguments, callee and such? > > I wonder (and will try to measure based on sunspider and other test content) > how many functions occur like function f(x) { return x +2; } or even fully > constant. Is there any classification of the complexity of functions at parse > time?

1 0

Generating bytecode documentation from Interpreter.cpp
by Holger Freyther 21 Sep '10

21 Sep '10

Hi all, it seems that squirrelfish-bytecode.html is out of sync with the reality again. A simple grep on the DEFINE_OPCODE (and removing some) shows that we have about 136 opcodes, searching for "<h2><code>" in the HTML only reveals 78 documented opcodes. I have tried to find a missing one (to support my hypothesis) and "op_eq_null" seems to be among the undocumented ones. Is there any interest to generate the documentation from the switch/case of the Interpreter.cpp? I think I could come up with a small python script to do that. comments?

2 2