Hello Alessandro, On Thu, Jun 12, 2025 at 12:45:15PM +0200, Alessandro Vesely wrote: > On Wed 11/Jun/2025 22:55:04 +0200 Uwe Kleine-König wrote: > > On Tue, Jun 10, 2025 at 09:07:34AM +0200, Alessandro Vesely wrote: > > > On resume, after reading back all saved memory and after displaying the > > > "Suspending console(s) (use no_console_suspend to debug)" message, > > > instead of > > > returning to the previous screen, it stay there and does not seem to > > > react. > > > Switching to and from virtual console, ctrl-alt-f1 + ctrl-alt-f7, results > > > in a > > > damaged screen where only the upper left corner is correct while the rest > > > replicates whatever was on the previous pixel. > > > > There is a (maybe) similar bug report at > > https://bugs.debian.org/1106668. > > > This seems different from bug 1106668. When the "Suspending console(s)" > message is displayed, besides ctrl-alt-fX, Caps and Num Locks' leds work > properly. The network works, I can login with ssh. > > > > I asked the reporter of that bug to check for a BIOS update > > My BIOS doesn't seem to be overly old: > > Vendor: American Megatrends Inc. > Version: 3211 > Release Date: 08/10/2021 > > I couldn't find which/ where more recent versions exist. > > > > and if that doesn't help follow > > https://docs.kernel.org/power/basic-pm-debugging.html to debug the > > issue. That would be helpful here, too. So can you please check these > > and follow up with your findings? > > > All three modes, reboot, platform and shutdown, fail in the same way > described above. That is, the screen is unusable after a circuitous > restore. > > The freezer test works fine. > > All the other tests, devices, platform, processors and core, resume to a > black screen. Blindly typed commands are accepted and executed. The > sequence ctrl-alt-f1 + ctrl-alt-f7 (which I called "circuitous") restores > the screen to a normal state. The test_resume option, after resetting > pm_test to none, behaves the same way. > > Probably the system works also after a real restore, except that the screen > is so broken as to make it unusable.
We discussed this bug during our weekly team meeting and with your results we have no idea how to address this. Our best recommendation is to test if this is an issue on a later kernel and then take this to upstream. So can you please check if that problem also happens with 6.12 or better still on 6.15, please? (Note that 6.12.31 has another amdgpu bug, so please make sure to test at least 6.12.32 (that is, take the kernel package from testing and not backports, or wait until backports are updated accordingly); experimental has 6.15.) The right place to report the problem upstream then is https://gitlab.freedesktop.org/drm/amd/-/issues. Given that there will probably some things to test in reply to that, it would be great if you could care for that report eliminating the kernel team as middle men. If so please mark this bug as forwarded with a link to your report (or reply with that info if you don't know how to do that). Best regards Uwe
signature.asc
Description: PGP signature
