Thread View: gnu.gcc.bug
1 messages
1 total messages
Started by Jacques Gelinas
Sat, 12 Sep 1992 05:14
gcc 2.2.2 on hpux generate bad assembly code
Author: Jacques Gelinas
Date: Sat, 12 Sep 1992 05:14
Date: Sat, 12 Sep 1992 05:14
45 lines
1835 bytes
1835 bytes
I have built GCC 2.2.2 on HPUX 8.02 on a HP9000-827. I have
managed to get a clean compile (stage1 stage2 stage3 and compare)
by playing some game. I have found two bugs in the assembly
code generated by the compiler.
The following program crashes with an illegal instruction trap:
int main()
{
float a=0;
a = a + 1;
}
The module enquire.c which generate float.h crash this way.
On the other hand, change float for double and all goes right.
So I rebuild the compiler, adding on the command line
"-Dfloat=double". Then I got a clean compile.
I start to compile a large CAD project, thinking that this
bug could be solve later (converting every float to double).
Doing so, after compiling a lot of code, I got another
problem. This time, it was flag by the assembler. It is related
to the number of parameter when generating the beginning of a function.
It generates and .export assembly statement:
Argument 1 or 3 in FARG upper
- lookahead = ARGW4=FR,RTNVAL=GR
The problem seems to happen with more than 4 parameters and
a combination of double and integer parameter.
I have ordered an assembly language reference manual. Maybe someone
have seen these problems. Any suggestion ?
Does a PA-RISC 1.1 can do single precision computation directly ?
Thanks in advance ?
--
--------------------------------------------------------
Jacques Gelinas (jack@solucor.uucp)
Today it's my opinion
Thread Navigation
This is a paginated view of messages in the thread with full content displayed inline.
Messages are displayed in chronological order, with the original post highlighted in green.
Use pagination controls to navigate through all messages in large threads.
Back to All Threads