dotgnu-libjit
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Libjit-developers] Re: libjit changes for archs with native ints (point


From: Klaus Treichel
Subject: [Libjit-developers] Re: libjit changes for archs with native ints (pointer sizes) greater than 32 bit
Date: Fri, 04 Jan 2008 12:35:07 +0100

Hi Aleksey,

from what i see after a third check the conversions are applied after
the types are normalized. This means jit_nint is normalized to jit_int
on 32bit and jit_long on 64 bit archs.
So the conversions shouldn't make any difference depending on the size
of native int.

Cheers, Klaus

Am Donnerstag, den 03.01.2008, 23:34 +0600 schrieb Aleksey Demakov:
> Hi Klaus,
> 
> > On making the interpreter running the pnet hello sample on x86_64 i
> > stepped over some code wrapped in #if JIT_NATIVE_INT32 or #if
> > JIT_NATIVE_INT64 in jit-insn.c that confuses me.
> >
> > So i have one question:
> >
> > The type jit_int is always 4 bytes in size. So why is there this
> > conditional compilation?
> 
> These #ifs are for the target platform's native int not for
> jit_int.
> 
> > An other issue is that pointers (native int) are longs on 64 bit archs.
> > Is it correct to set the is_nint_constant in
> > jit_value_create_long_constant?
> >
> 
> What problem does it solve? I see no need for this change.
> 
> >
> > Doing some of the changes made pnet's hello sample run with libjit
> > interpreter on x86_64.
> >
> 
> That's strange that it made a difference in your setup. These
> changes do not seem to fix any real problem for me.
> 
> Regards,
> Aleksey

Attachment: signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil


reply via email to

[Prev in Thread] Current Thread [Next in Thread]