As development moves on the big bugs disappear and the small ones start scurrying around. Especially on the IOP. These are hard to track down if it isn't your code, so... do us a favor and download the package at the bottom of this message and use it to build your very own binutils-2.14 and gcc-3.2.2 toolchain. Then, test your code. Test other code. Try to break it. Then the most important part of all: POST YOUR BUGS FOR ALL TO SEE.
Together we can stomp on all of the bugs and pave the way to the magic world of gcc-3.4 and beyond...
http://www.oopo.net/consoledev/toolchain-beta.tar.gz
We need YOU to help test the new toolchain.
-
mountainstorm
- Posts: 21
- Joined: Wed Apr 07, 2004 2:59 am
The address of ps2client in the readme.txt is wrong, it should be
http://www.oopo.net/consoledev/files/ps ... 0.0.tar.gz
MountainStorm
http://www.oopo.net/consoledev/files/ps ... 0.0.tar.gz
MountainStorm
www.oopo.net not responding
Interesting in testing the new toolchain, but www.oopo.net does not seem to be responding at the moment..
Re: We need YOU to help test the new toolchain.
The assembly patch by ryani seems to have been forgotten somewhere (no comments, not in current patches from what I can tell)
http://www.mindspring.com/~foo/binutils ... 9ryani.txt
posted in
http://forums.ps2dev.org/viewtopic.php?t=80 (pixels thread)
http://www.mindspring.com/~foo/binutils ... 9ryani.txt
posted in
http://forums.ps2dev.org/viewtopic.php?t=80 (pixels thread)
-
Guest
This isn't so much testing of the new toolchain as it is testing
of the toolchain-beta.sh install script, a fantastic tool if there
ever was one.
I have run it on both Mac OS X 10.3.x and FreeBSD 5.2.x.
Ran into two minor problems, one suffered by both platforms,
one by only one.
1. Both OSX and FreeBSD were initially lacking of wget. Had to
download and install it. No big deal, but both also then had
problems where IPV6 was enabled and wget choked it first
tries IPV6 to talk to the server then switches to IPV4. For
some reason, the switch to IPV4 just never works out. On
the MAC the easy solution was to disable IPV6, but I couldn't
figure out how to do it on FreeBSD so used --bind-address to
wget with the local IPV4 address and it got by.
2. FreeBSD apparently doesn't come with GNU make, and GNU
make is now pretty much a requirement to compile the GNU
toolchains. Its easy to install GNU make on FreeBSD, but I
wanted to mention it for anyone else who wonders why the
toolchain install ends up so broken - install GNU make and
start all over, things will be fine.
One other minor problem, I originally had problems getting newlib
from redhat.com and had to find another mirror. Its too bad wget
can't (or maybe it can?) be given alternate sources, because it
requires minor hacking to get get past.
Other than that...new toolchain has been fantastic. Oh there are minor
makefile tweaks required here and there...but no biggie.
Gorim
of the toolchain-beta.sh install script, a fantastic tool if there
ever was one.
I have run it on both Mac OS X 10.3.x and FreeBSD 5.2.x.
Ran into two minor problems, one suffered by both platforms,
one by only one.
1. Both OSX and FreeBSD were initially lacking of wget. Had to
download and install it. No big deal, but both also then had
problems where IPV6 was enabled and wget choked it first
tries IPV6 to talk to the server then switches to IPV4. For
some reason, the switch to IPV4 just never works out. On
the MAC the easy solution was to disable IPV6, but I couldn't
figure out how to do it on FreeBSD so used --bind-address to
wget with the local IPV4 address and it got by.
2. FreeBSD apparently doesn't come with GNU make, and GNU
make is now pretty much a requirement to compile the GNU
toolchains. Its easy to install GNU make on FreeBSD, but I
wanted to mention it for anyone else who wonders why the
toolchain install ends up so broken - install GNU make and
start all over, things will be fine.
One other minor problem, I originally had problems getting newlib
from redhat.com and had to find another mirror. Its too bad wget
can't (or maybe it can?) be given alternate sources, because it
requires minor hacking to get get past.
Other than that...new toolchain has been fantastic. Oh there are minor
makefile tweaks required here and there...but no biggie.
Gorim
Blackdroid actually found several bugs in my binutils's patch, rendering several vu instructions unworkable.
pixel: A mischievous magical spirit associated with screen displays. The computer industry has frequently borrowed from mythology. Witness the sprites in computer graphics, the demons in artificial intelligence and the trolls in the marketing department.
-
blackdroid
- Posts: 564
- Joined: Sat Jan 17, 2004 10:22 am
- Location: Sweden
- Contact:
vu macro, aka cop2 instructions.. basically vdiv, vrsqrt, vrinit, vrxor if wanting to use 'w' as destination. I have 2 other bugs im looking at atm. one is pretty insignificant for most ppl since its just a rarely used gas feature that gives an assert.
the other has to do with alignment and -O2.
the other has to do with alignment and -O2.
Kung VU
The latest update to the beta script checks for gmake, falling back to make if none is found. Or it should. :)
As for the IPv6 stuff, I can't say I've run into that problem under OSX, or even NetBSD. This problem may also be related to your attempts to get newlib from redhat. Anyway, if you manually grab the archives and put them in the same directory as the script wget should see them and skip the download. I suppose I could get the script to check instead for a more proper solution...
bd: Dude! Keep up the work. :)
As for the IPv6 stuff, I can't say I've run into that problem under OSX, or even NetBSD. This problem may also be related to your attempts to get newlib from redhat. Anyway, if you manually grab the archives and put them in the same directory as the script wget should see them and skip the download. I suppose I could get the script to check instead for a more proper solution...
bd: Dude! Keep up the work. :)