Bug 26237

Summary:	32-bit kernel-desktop 5.5.4-1 has video artifacts on old Radeon GPU
Product:	Mageia	Reporter:	Thomas Andrews <andrewsfarm>
Component:	RPM Packages	Assignee:	Kernel and Drivers maintainers <kernel>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	Normal	CC:	tmb
Version:	7
Target Milestone:	---
Hardware:	i586
OS:	Linux
Whiteboard:
Source RPM:	kernel-5.5.4-1.mga7.src.rpm	CVE:
Status comment:

Description Thomas Andrews 2020-02-22 15:58:58 CET

Description of problem: After updating to kernel-desktop 5.5.4-1 and related packages, and rebooting, visual artifacts are seen that make the desktop completely unusable. The artifacts affect the panel and desktop icons, but not the background image. It appears to be an error in re-drawing, as if some lines were not being rendered correctly.

This problem was observed in early versions of the 5.4 series kernel-desktop, but was corrected in later ones. If the user boots into kernel-desktop 5.4.17, the symptoms disappear.

Affected hardware: Dell Inspiron 5100, 32-bit P4 processor, 2GB of RAM, Radeon Mobility 7500 (RV200) graphics, and an old Atheros wifi card. When the problem was affecting the 5.4 series kernel, other old Radeon GPUs were affected, as well.


Version-Release number of selected component (if applicable):

kernel-desktop-5.5.4-1.mga7-1-1.mga7.i586.rpm

How reproducible:

Always

Steps to Reproduce:
1. Update/install kernel 5.5.4-1 on affected hardware
2. Reboot
3.

Thomas Backlund 2020-02-22 16:09:42 CET

Assignee: bugsquad => kernel
CC: (none) => tmb

Comment 1 Thomas Backlund 2020-02-22 17:19:19 CET

Ok, looks like upstream have not any ideas yet as there is a similar report on LKML, but no resolution besides switching of HIGHMEM support, which I wont do in a distro kernel.

I've pinged the thread on LKML but it looks like for now I will have to revert the same patches as I did in 5.4 series kernels...

Comment 2 Thomas Andrews 2020-02-22 17:34:49 CET

A random thought... I'm not really in the loop on this sort of thing, but it seems to me that I started hearing about the i915 freeze-up problems just about the time you applied those patches to the 5.4 series kernels.

Is there any way that could be related, or is it just a coincidence?

Comment 3 Thomas Backlund 2020-02-22 17:43:51 CET

just a coincidence.

the radeon issues are specific to the radeon driver and dma mapping

the i915 issues are related to hardware hangs because of driver/firmware issues

and the radeon issues started with 5.4 series, whereas for some the i915 issues started to show up at 5.1 series and then depending on hw at 5.2, 5.3 or 5.4 series...

And the security fixes that landed in i915 driver exposed even more issues, and losing power save modes...

Comment 4 Thomas Andrews 2020-02-22 18:13:57 CET

OK, thanks for explaining it in words that a layman can understand.

Comment 5 Thomas Backlund 2020-02-24 18:50:06 CET

There is now a kernel-5.5.6-1.mga7 in updates_testing to try

Comment 6 Thomas Andrews 2020-02-24 20:50:00 CET

Installed cpupower and kernel-desktop on my Inspiron 5100. Also updated kernel-firmware-nonfree, even though it probably wouldn't affect anything on this hardware.

Rebooted, and all is well. Graphics artifacts are gone again.

Thank you, Thomas!

Comment 7 Thomas Andrews 2020-02-26 22:29:21 CET

Changing status to Resolved.

Resolution: (none) => FIXED
Status: NEW => RESOLVED

Comment 8 Thomas Backlund 2020-02-29 15:00:31 CET

An update for this issue has been pushed to the Mageia Updates repository.

https://advisories.mageia.org/MGASA-2020-0110.html