To: None <firstname.lastname@example.org>
From: Dirk H. Hohndel <email@example.com>
Date: 11/25/1998 16:41:13
XFree86-3.3.3 has been released
The XFree86 Project, Inc., is proud to announce its latest release,
This is the latest in our series of "final XFree86-3.3.x release"
Most of our work is focused on XFree86-4.0 these days, but the amazing
shelftime of graphics hardware makes another "old design" release
Here's an excerpt from the RELNOTES:
1. XFree86 and X11R6.3
XFree86 releases starting with the 3.2A beta are based on the X Consortium's
X11R6.3 (the final release from the X Consortium).
R6.3 is an update to R6.1, and is intended to be compatible with R6.1 and R6 at
the source and protocol levels. Binaries should be upward-compatible. X11R6.3
includes some new Xserver extensions: SECURITY, XC-APPGROUP, XpExtension (print
extension), and an updated, and standardised version of LBX. X11R6.3 also has
new standards, including RX (X Remote Execution MIME type), and a proxy manage-
ment protocol. X11R6.3 includes support for gzipped fonts.
R6.1 is an update to R6, and is intended to be compatible with R6 at the source
and protocol levels. Binaries should be upward-compatible. X11R6.1 includes
some new Xserver extensions: DOUBLE-BUFFER, XKEYBOARD and RECORD.
What about R6.2? X11R6.2 is the name given to a subset of X11R6.3, which has
only the print extension and the Xlib implementation of vertical writing and
user-defined character support in addition to those features included in R6.1.
In September 1998 The Open Group changed the copyright of X11R6.4 from the non-
free copyright used for the initial release of X11R6.4 back to the old free X
Consortium style copyright. Given the fact that our main development focus is
on XFree86-4.0 and that the 3.3.3 release was intended mostly to fix some bugs
and get out new drivers to the public, we opted not to include X11R6.4 in
XFree86-4.0 will be X11R6.4 based.
3. OS issues
Always check the OS specific README files for special requirements or caveats.
Users running Linux should note that Elf is now the only binary type supported
for Linux OSs. This means that binaries for ix86/a.out and AXP/ECOFF are not
available with this release.
Users running FreeBSD 3.0 should note that only ELF binaries are provided with
this release. a.out shared libraries are also included for compatibility pur-
4. What's new in 3.3.3?
4.1 Security fixes
o Several buffer overrun problem discovered since the release of
XFree86-3.3.2 have been fixed
o Sanity checks on DISPLAY variable
o Attempt to stop X connection hijacking (sticky bit for /tmp/.X11-unix)
Note that this is only a short-term partial solution, and it is doesn't
help at all for some SYSV based OSs (like Solaris 2.x).
4.2 Bug fixes
o Fix a serious LBX bug using uninitialized variables.
o Fix some Xlib bugs that cause problems when using XKB in some locales
(like latin2), add support for iso8859-15, and include a couple of basic
fonts for iso8859-15
o Fix xf86config to handle more than 10 modes and to be prepared for XFCom /
o Lots of xterm changes, see xterm.log.html in the sources.
o Fix problems with high dot clocks in high color depths on Riva128.
o Fix problem in the S3 drivers with disabled onboard S3 chips when using S3
o Fix problems with Cirrus 5480 at high resolutions and jitter that appeared
with the 546x's using the BitBLT engine.
o Fix clock limits in some cases in Tseng driver.
o Fix some lockups with ViRGE chips.
o Improved timing calculations for video FIFO in the Mach64 X server.
o Fixed bug in font rendering code in the Mach64 X server.
o Fixed VGA font restoration bug when exiting the Mach64 X server.
o Several XF68_FBDev fixes.
o Fix wrong clock limits for S3 Trio64V+.
o Fix some generic rendering errors in cfb and vga code.
o Fix text restore problems and improve high res 32bpp modes in MGA driver;
fix 24bpp and 32bpp display problems; disable probing for memory on some
MGA chipsets; fix maximum blit size; fix sync on green for Mystique.
o Fix problems with Xnest crashing with too many visuals.
o Fixes for 64bit architectures.
o Fix cursor bug in S3V server.
o Fixes for memory probing, max dotclock probing and DPMS display off on C&T
o Fix LCD detection for CLGD755x and the double mouse bug and the blanking
bug in the cirrus driver.
o Fix some problems with -quiet flag (where some variables stayed uninitial-
o SuperProbe can detect C&T HiQV chips now, with an exception in the case of
No-PCI bus connected. The "-no_bios" option of SuperProbe solve this situ-
o The C&T chipsets now use software cursors by default to avoid a number of
minor problems in certain circumstances. Hardware cursors can still be
used by adding the "hw_cursor" option to XF86Config.
o EGC server now works on Linux/98. XF98Setup also works with it.
o Fix VT switch problem with MGA server on Linux/98.
4.3 New Features
o New driver for Cyrix MediaGX based motherboards.
o New driver for Rendition V1000 and V2x00 chipsets (not accelerated).
o New driver for Weitek P9100 based cards.
o New driver for SiS 5597/98 and SiS 6326; treat SiS 6215 and 6225 as 6205.
o New server for 3Dlabs based cards using GLINT 500TX and MX (with IBM RAM-
DAC), Permedia (with IBM RAMDAC), Permedia 2 and Permedia 2v.
o Support for the Matrox G100 and G200 based cards to the MGA driver.
o Support for C&T 69000 and 32bpp on 65550 and later.
o Support for NeoMagic notebook chipsets.
o Support for EPSON SPC8110.
o Support for NVidia Riva TNT.
o Acceleration for Trident Image975, Image985, Cyber9397, Cyber9388.
o Support for the new ATI Rage Pro, VT4 and Rage IIC based cards has been
o 24-plane TGA support.
o Config support to build XFree86 on Linux with DECnet transport.
o Support to build XFree86 for FreeBSD/ELF.
o Support for vesafb on Linux/x86.
o LynxOS 3.0 support.
o Updates to SuperProbe.
o New XInput drivers for AceCad ADVANCEDigitizer, MicroTouch TouchPen, SGI
o Add local font directory.
4.4 Known Problems
o There are problems with some Cirrus laptop chipsets (75xx). The driver
seems to work for some people, but not others. Until someone with the
appropriate hardware can look into this, these problems are unlikely to be
fixed. If you wish to work on this, please contact us. We don't need
testers, we need people willing and able to fix the problems.
o There are problems with some of the Trident laptop chipsets. The driver
seems to work in a limited way for some people, but not others. Until
someone with the appropriate hardware can look into this, these problems
are unlikely to be fixed. If you wish to work on this, please contact us.
We don't need testers, we need people willing and able to fix the prob-
o SuperProbe command fails to detect some newly supported chips. Currently,
the probing result with Xserver itself with appropriate setting in
XF86Config (and maybe option "-probeonly") can be more relied on than the
result from SuperProbe for newer chips. If you are interested in
improving the design of SuperProbe's code, let's come and join as the mem-
ber of the XFree86 ``developer team''.