I've been looking deeper into this problem, which was reportedly introduced by
my patch a while ago, in Linux kernel 2.6.17. I don't doubt the reports, but
I'd like to replicate the problem as seen on the x86 boxes as reported in kernel
bug 6720 and a few others directly to me, so I can verify a fix (if/when I can
create one).
So, I installed SLES 10 on an old x86 box, with a PCI G450 (actually an IBM
GXT135P). I verified that my patch is _not_ included in that kernel
(2.6.16.21). However, I see jittery DVI-D output at all times. I've tried
640x480@60Hz, 1280x1024@60Hz, different "vesa" mode parameters (0x11b, 0x1bb),
XServer "-fbbpp 32". I usually end up losing sync with the monitor after a few
minutes.
I don't have much experience with the G450 on x86, but I'm quite surprised. Has
anyone else observed this?
--
Regards,
Paul Clarke
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Paul A. Clarke wrote:
I've been looking deeper into this problem, which was reportedly
introduced by my patch a while ago, in Linux kernel 2.6.17. I don't
doubt the reports, but I'd like to replicate the problem as seen on the
x86 boxes as reported in kernel bug 6720 and a few others directly to
me, so I can verify a fix (if/when I can create one).
So, I installed SLES 10 on an old x86 box, with a PCI G450 (actually an
IBM GXT135P). I verified that my patch is _not_ included in that kernel
(2.6.16.21). However, I see jittery DVI-D output at all times. I've
tried 640x480@60Hz, 1280x1024@60Hz, different "vesa" mode parameters
(0x11b, 0x1bb), XServer "-fbbpp 32". I usually end up losing sync with
the monitor after a few minutes.
I don't have much experience with the G450 on x86, but I'm quite
surprised. Has anyone else observed this?
I believe that problem is not x86 vs. ppc, but type of monitor you have
connected to the box.
Perhaps try walkdvi
(http://platan.vc.cvut.cz/ftp/pub/linux/matrox-latest/walkdvi.tar.gz, it will
probably need some fixes for big endian) with different XDVICLKCTL values. Code
to set XPANMODE (in g450_set_plls) was written based on experimenting with
couple of devices I had, so maybe right thing is not modifying XDVICLKCTL
register, but instead teaching PLL code to try different XPANMODE values if PLL
does not lock in reasonable time.
Petr
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642