1

This problem happens on a laptop with AMD e-350 APU and built-in radeon HD 6310 graphics. It runs Slackware64 version 14.0. On the console (VTs), KMS is used with the radeon driver.

The problem occurs when using a 3.10.* series kernel. If console screen blanking is enabled (e.g., setterm -blank or kernel parameter consoleblank=) then when the screen blanks, it momentarily becomes black as it should, but after a second or less turns entirely white and remains that way until it unblanks. I've experimented and searched the web without any success at solving this.

I will add that the screen blanks properly if either (a) KMS is disabled and the console video is configured with vga=, or (b) a 3.2.* series kernel is used with KMS. These possibilities are less desirable for my use, for various reasons. I'd like to get it working right with 3.10.* kernels.

I'd welcome any suggestions. I can also give more info if needed.

  • I searched some more and found this thread, which indicates that the problem can be fixed with the radeon parameter backlight=0, which is available on 3.18.* kernels and newer. – serval Mar 15 '16 at 3:38
  • See also this question, which has the same answer. – calavicci - GoFundMonica 2 days ago
1

I searched some more and found this thread, which indicates that the problem can be fixed with the radeon parameter backlight=0, which is available on 3.18.* kernels and newer.

I happened to have 3.18.7 to try, and that worked perfectly with backlight0. However there was a new problem: this kernel broke my firewall, which required a re-config and recompile to fix. Given the need to recompile, I updated the source to the latest 3.18.28 before recompiling. With this, the firewall was fixed, but the console blanking fix no longer worked. Now, with backlight=0 the whole console goes permanently dark the moment that radeon is loaded. No way to get it back except reboot. If I try backlight=1 instead, the console blanks OK the first time, but blanks to white subsequently (as described in the above-linked thread).

So, this has morphed into a related problem that needs more investigation.

1

I've done some more investigation. First I looked at all the incremental patches that lead from 3.18.7 (blanking worked ok with backlight=0) to 3.18.28 (broken). The most suspicious change was in the patch from .22 to .23.

So I compiled 3.18.22 and it worked with backlight=0. Then moving to 3.18.23, it was indeed broken: backlight=0 leads to a permanently black screen again. Then I removed the two suspicious lines from 3.18.23's code. They occur in file drivers/gpu/drm/radeon/atombios_encoders.c and are identical calls:

atombios_set_backlight_level(radeon_encoder, dig->backlight_level);

If I comment these calls out, and recompile, it works OK. I suspect only one of these calls may need to be removed on any given hardware having this problem.

Finally, I want to note that I tried a newer kernel, ver 4.4.5 from Slackware64 -current. This works well with backlight=0, no need to hack anything.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.