Bug#317756: Image defects with 2.6.14/atyfb/Sunblade 100 (fwd)
Thanks, Luis. I'm forwarding your information to the bug so that it'll not
get lost. I'll also write to debian-sparc list, a few people have reported
that corruption there recently, so they will probably willing to test it
out.
Best regards,
Jurij Smakov jurij@wooyd.org
Key: http://www.wooyd.org/pgpkey/ KeyID: C99E03CC
---------- Forwarded message ----------
Date: Wed, 28 Dec 2005 20:16:23 -0500
From: Luis F. Ortiz <lfo@Polyad.Org>
To: jurij@wooyd.org
Cc: Andrey Chernomyrdin <andrey.debian-sparc@sats.spb.ru>, jim@amarooas.com.au,
Dave Love <d.love@dl.ac.uk>
Subject: Re: Image defects with 2.6.14/atyfb/Sunblade 100
On Nov 12, 2005, at 12:51 AM, Jurij Smakov wrote:
Recently Luigi Gangitano has pointed out a linux-fbdev-devel thread [0]
in which progress was achieved in fixing the atyfb bug resulting in console
and X display corruption on SunBlade 100, filed as #321200 in the Debian bug
tracking system [1]. The patch mentioned in the thread (which we plan to
include in the next Debian kernel package release) is attached. While the
situation with this patch is greatly improved, users testing the new 2.6.14
kernel packages have reported [2,3] that there is still some minor screen
corruption in a form of red dots or lines. A screenshot (provided by Andrey
Chernomyrdin) can be viewed at [4]. We'd appreciate any ideas which could
lead to the resolution of this problem.
Jurij:
I happened to be (forced into) upgrading my system at home from 2.4.27
to 2.6.14 today and ran into the problem that
you described. I did a review of the 2.4 and 2.6 trees and a couple of tests
and ended up with the following fix:
lfo@henkle:/usr/src/linux-2.6/drivers/video/aty$ cg-diff atyfb_base.c
diff --git a/drivers/video/aty/atyfb_base.c b/drivers/video/aty/atyfb_base.c
old mode 100644
new mode 100755
index 037fe9d..f5b8f45
--- a/drivers/video/aty/atyfb_base.c
+++ b/drivers/video/aty/atyfb_base.c
@@ -404,7 +405,7 @@ static struct {
{ PCI_CHIP_MACH64GM, "3D RAGE XL (Mach64 GM, AGP)", 230, 83, 63,
ATI_CHIP_264XL },
{ PCI_CHIP_MACH64GN, "3D RAGE XL (Mach64 GN, AGP)", 230, 83, 63,
ATI_CHIP_264XL },
{ PCI_CHIP_MACH64GO, "3D RAGE XL (Mach64 GO, PCI-66/BGA)", 230, 83, 63,
ATI_CHIP_264XL },
- { PCI_CHIP_MACH64GR, "3D RAGE XL (Mach64 GR, PCI-33MHz)", 230, 83, 63,
ATI_CHIP_264XL },
+ { PCI_CHIP_MACH64GR, "3D RAGE XL (Mach64 GR, PCI-33MHz)", 235, 83, 63,
ATI_CHIP_264XL | M64F_SDRAM_MAGIC_PLL},
{ PCI_CHIP_MACH64GL, "3D RAGE XL (Mach64 GL, PCI)", 230, 83, 63,
ATI_CHIP_264XL },
{ PCI_CHIP_MACH64GS, "3D RAGE XL (Mach64 GS, PCI)", 230, 83, 63,
ATI_CHIP_264XL },
Note that it is the same as the patch you have, except that I also corrected
the PLL frequency from
230 to 235 Mhz. It looks like a typo when copying during the rewrite the aty
directory underwent.
Perhaps others can give this change a try and see if it improves their
problems.
It fixed both my FB console problems and my problems when starting Xfree 4.3 .
I've only given
it a few hours of testing at three different screen resolutions, so there are
plenty of opportunities
for other people to find different things wrong.
Note that this is a patch on
rsync://rsync.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git .
If someone lets me know this is looking good for them, I'll pass it along with
my other patches to
David Miller and let him push it into the official kernel.
Luis Ortiz
Reply to: