--- Begin Message ---
- From: Niels Langendorff <Niels dot Langendorff at chess dot nl>
- To: Jeroen Roose <Jeroen dot Roose at chess dot nl>
- Date: Fri, 08 Nov 2002 15:07:49 +0100
- Subject: gcc vraagje
- References: <3DCBAE79.9060707@chess.nl>
Hi,
I'm porting the Sun CVM 1.0.1 for the Arm platform, and it looks like
the Java Double (defined as a long long) from the C-stack is not
returned to the Java stack.
Looking through the man pages of gcc, I've read that large values are
returned in a structure, and a pointer to this structure is placed on
the C-stack. Not expecting this kind of behaviour, and not want to
modify the Arm assembly code, I was wondering if there is a compiler
option (or a workaround) for returning the long long value onto the
stack, instead of a pointer to it.
I'm using the following configuration:
gcc version 2.95.3-5 (cygwin special)
Linux 2.4.16-rmk2, NETWinder Floating Point Emulator V0.95
Best regards,
Niels Langendorff
begin:vcard
n:Langendorff;Niels
tel;fax:+31 (0)23 5 149 199
tel;work:+31 (0)23 5 149 123
x-mozilla-html:FALSE
url:www.chess.nl
org:Chess eT
version:2.1
email;internet:Niels.Langendorff@chess.nl
title:Project Engineer
adr;quoted-printable:;;Visit =0D=0AChess eT=0D=0ANieuwe Gracht 13=0D=0A2011 NB Haarlem=0D=0AThe Netherlands=0D=0A=0D=0AExpedition =0D=0AChess eT=0D=0AParklaan 20 =0D=0A2011 KW Haarlem=0D=0AThe Netherlands=0D=0A=0D=0APost=0D=0AChess eT=0D=0APostbus 5021=0D=0A2000 CA Haarlem=0D=0AThe Netherlands;Haarlem;;2011 NC;The Netherlands
fn:Niels Langendorff
end:vcard
--- End Message ---