Re: [Lightning] Porting GNU Smalltalk to lightning 2

[Paulo César Pereira de Andrade]

2014-10-26 19:17 GMT-02:00 Holger Hans Peter Freyther <address@hidden>:
> On Sun, Oct 26, 2014 at 04:14:15PM -0200, Paulo César Pereira de Andrade 
> wrote:
>
>>   It is awful :(
>
> Yes, specially with R^X.. I see. and I do prefer lightning dealing
> with PROT_EXEC, flushing the cache of the CPU, etc.

  Anyway, I think it is a good idea to provide a call to flush the
cache, when applicable, to match lightning 1

void
jit_flush_code(void *start, void *end);

prototype. In lightning 2, in most backends, it assumes gcc,
and linked to libgcc, and calls "__clear_cache(void*,void*);",
save on a few backends, where the libgcc __clear_cache does
not work.

>>   The easiest approach is to just use an variably sized structure,
>> and use jit_set_data, see "7.2 Alternate code buffer" in
>> http://www.gnu.org/software/lightning/manual/html_node/Customizations.html#Customizations
>> the example there uses mmap, but you may use malloc'ed
>> memory, no problem (and see the comments on how you can
>> even query the likely size of the buffer, and retry unitl it fits...),
>> just that lightning does leave you on your own about cache sync,
>> so, with plain memory from malloc, it likely will only work as is
>> on ix86 or x86_64, and maybe a few others, while some are very
>> "hard" to sync.
>
> right, it is not where I want to go. The invalidation is very
> noticable for interactive workloads (gst-browser compiles a method,
> adds it to the method dictionary and then executes it) so I wonder
> what other options we have. But I think you are right and the hash
> you added is the right way to move forward right now.
>
>
>>   I may work on a new api to ensure the memory is synch'ed
>> tough. I hope to have lightning 2.0.6 released when it can
>> drive GNU Smalltalk :)
>
> I want to give it a try on ARM as well. Hopefully this week.

  I believe it is very likely to "just" work.

  For 64 bit, my patch is a complete hack. Need to write a
proper division code if there isn't an overflow fallback (or an
assertion if it must always exit). Probably just shift right
tag bits, divide, shift left tag bits...

Thanks,
Paulo