This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re[2]: gcc --target=i386-realmode soon ?
- To: Ian Lance Taylor <ian at zembu dot com>, egcs at egcs dot cygnus dot com
- Subject: Re[2]: gcc --target=i386-realmode soon ?
- From: Etienne Lorrain <etienne dot lorrain at ibm dot net>
- Date: Thu, 15 Jul 1999 11:51:42 +0100
- CC: eliot at mgm dot mit dot edu, bug-gnu-utils at gnu dot org
- Organization: Protection of the "jnp" instruction league
- References: <378CAFBF.AA53B2E1@ibm.net> <19990714175523.12536.qmail@daffy.airs.com>
Ian Lance Taylor wrote:
>
> You may already be aware of this, but note that the development
> version of gas includes much better real mode support than the code in
> the 2.9.1 release. See
> http://sourceware.cygnus.com/binutils
> for the binutils development sources.
>
Hello,
I am already using a (few month old) development release
of GAS, because of the buggy hexadecimal encoding of:
mov %ax,2(%bp)
mov %ax,2(%di)
mov %ax,2(%bp)
Those generate the same hexadecimal with binutil-2.9.1 .
But in fact I did not do the paper work needed to
submit a patch to GNU projects (FSF), so I have just
described the simple changes. Anyway the addr32 part
is just a "dirty patch", which works but should
probably be more integrated to the algorithm: you
already detect this case, but flag it as an error,
in fact it is perfectly working on every PC it tried,
from AMD 386 to AMD K6 350 or Pentium II 400.
I can try to generate an up-to-date patch for GAS,
but in fact my VESA 2 / big (4Gb) flat real mode bug
is still bothering me... mostly when VESA 1.x works.
Thanks,
Etienne.