http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c1 Andreas Stenglein changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |Andreas.Stenglein@mailbox.o | |rg --- Comment #1 from Andreas Stenglein --- I have a similar issue with Pentium 4 and Radeon RV370/RV380: kernel-pae-4.18.15-1.2.i686 works kernel-pae-4.19.x doesn't work: triggers reboot, almost immediately at startup Workarounds: kernel option: nomodeset (system boots up, but Xorg doesn't work) kernel option: nopti (system boots and is usable, but no protection against meltdown) I also tried default kernel 4.19.x: works. Note: The graphics chip is reported either as RV370 (lspci) or RV380 (dmesg) kernel git bisect from 4.18 to 4.19: 7757d607c6b31867777de42e1fb0210b9c5d8b70 is the first bad commit commit 7757d607c6b31867777de42e1fb0210b9c5d8b70 Author: Joerg Roedel Date: Wed Jul 18 11:41:14 2018 +0200 x86/pti: Allow CONFIG_PAGE_TABLE_ISOLATION for x86_32 Allow PTI to be compiled on x86_32. Signed-off-by: Joerg Roedel Signed-off-by: Thomas Gleixner Tested-by: Pavel Machek Cc: "H . Peter Anvin" Cc: linux-mm@kvack.org Cc: Linus Torvalds Cc: Andy Lutomirski Cc: Dave Hansen Cc: Josh Poimboeuf Cc: Juergen Gross Cc: Peter Zijlstra Cc: Borislav Petkov Cc: Jiri Kosina Cc: Boris Ostrovsky Cc: Brian Gerst Cc: David Laight Cc: Denys Vlasenko Cc: Eduardo Valentin Cc: Greg KH Cc: Will Deacon Cc: aliguori@amazon.com Cc: daniel.gruss@iaik.tugraz.at Cc: hughd@google.com Cc: keescook@google.com Cc: Andrea Arcangeli Cc: Waiman Long Cc: "David H . Gutteridge" Cc: joro@8bytes.org Link: https://lkml.kernel.org/r/1531906876-13451-38-git-send-email-joro@8bytes.org :040000 040000 dbab9a897d534d7b14f900f0c6779b6848833892 f0674017544bc95fafa431d1e638f994eca37b51 M security -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c3 --- Comment #3 from Andreas Stenglein --- Created attachment 793389 --> http://bugzilla.opensuse.org/attachment.cgi?id=793389&action=edit dmesg of kernel with option nopti set -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c5 --- Comment #5 from Andreas Stenglein --- Created attachment 793391 --> http://bugzilla.opensuse.org/attachment.cgi?id=793391&action=edit iomem as root -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c7 --- Comment #7 from Andreas Stenglein --- Created attachment 793393 --> http://bugzilla.opensuse.org/attachment.cgi?id=793393&action=edit lcpci -vvv as root -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c9 --- Comment #9 from Andreas Stenglein --- I have to admit that during bisecting I noticed 2 internal compiler errors that just went away after typing make / make bzImage / make modules again. And I noticed some freezes during compilation. Otherwise the machine seems fine; I didn't notice similar problems before. I did the bisecting/compiling with kernel_4.18.15-1-pae -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 Andreas Stenglein changed: What |Removed |Added ---------------------------------------------------------------------------- Priority|P5 - None |P3 - Medium -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c12 --- Comment #12 from Andreas Stenglein --- Created attachment 793426 --> http://bugzilla.opensuse.org/attachment.cgi?id=793426&action=edit oops #1, via serial connection Linux version 4.19.12-1.gc634493-pae (geeko@buildhost) (gcc version 8.2.1 20181108 [gcc-8-branch revision 265914] (SUSE Linux)) #1 SMP PREEMPT Sat Dec 22 08:59:36 UTC 2018 (c634493) -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c14 --- Comment #14 from Andreas Stenglein --- Created attachment 793428 --> http://bugzilla.opensuse.org/attachment.cgi?id=793428&action=edit oops #3 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c16 --- Comment #16 from Andreas Stenglein --- Created attachment 793430 --> http://bugzilla.opensuse.org/attachment.cgi?id=793430&action=edit oops #5 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c18 --- Comment #18 from Andreas Stenglein --- Created attachment 793432 --> http://bugzilla.opensuse.org/attachment.cgi?id=793432&action=edit oops #7 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c20 --- Comment #20 from Andreas Stenglein --- workaround: disabling Hyper Threading in BIOS helps (only 1 CPU available) plymouth.enable=0 at kernel command line doesn't help -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c21 --- Comment #21 from Felix Miata --- 4.19.11 pae boots to multi-user, but there's no keyboard trying to login. CAD produces many screens of code dumps, then lockup. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c23 --- Comment #23 from Andreas Stenglein --- Bug still present with kernel-pae-4.20.1-1.1.g5978cc8.i686 kernel-pae-5.0.rc1-1.1.ga56baa9.i686 here it only happens with hyperthreading enabled (>1 cpu) && graphical boot / vesafb && pti && kernel based modesetting During my tests once one of the last lines shown before lockup was: Console: switching to colour frame buffer device 160x64 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c25 --- Comment #25 from Andreas Stenglein --- bug exists also in: kernel-pae-5.0.rc2-1.1.g879eb5c.i686 kernel-pae-4.20.2-3.1.g036c5c2.i686 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c27 --- Comment #27 from Andreas Stenglein --- bug exists also in: kernel-pae-5.0.rc4-1.1.g8e6abff.i686 kernel-pae-4.20.5-3.1.g0a5bbbc.i686 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 Andreas Stenglein changed: What |Removed |Added ---------------------------------------------------------------------------- Summary|32bit x86 4.19 pae and |32bit x86 4.19 4.20 5.0-rc |vanilla kernels randomly |pae and vanilla kernels |lockup or trigger reboot |randomly lockup or trigger | |reboot -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c30 --- Comment #30 from Andreas Stenglein --- also affected: kernel-pae-5.0.rc7-1.1.gb094e66.i686 kernel-pae-4.20.10-2.1.g4c06520.i686 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c32 --- Comment #32 from Felix Miata --- Virtually all MBR Gnu/Linux "booting" here is via openSUSE 13.1's grub-0.97.194. Plymouth is either not installed, or cmdline option plymouth.enable=0 is applied. (All UEFI booting here is also non-graphical.) # cat /proc/cmdline root=LABEL=f28p22 ro audit=0 ipv6.disable=1 net.ifnames=0 selinux=0 plymouth.enable=0 noresume vga=791 video=1440x900@60 3 # uname -a Linux gx280 4.20.6-100.fc28.i686 #1 SMP Thu Jan 31 15:54:55 UTC 2019 i686 i686 i386 GNU/Linux # dmesg | egrep -i 'pti|vesafb' [ 0.000000] DMI: Dell Inc. OptiPlex GX280 /0G5611, BIOS A08 03/03/2006 [ 2.242347] vesafb: mode is 1024x768x16, linelength=2048, pages=4 [ 2.242357] vesafb: scrolling: redraw [ 2.242364] vesafb: Truecolor: size=0:5:6:5, shift=0:11:5:0 [ 2.242389] vesafb: framebuffer at 0xc0000000, mapped to 0x(ptrval), using 3072k, total 7872k intel_rng: RNG, try using the 'no_fwh_detect' option. [ 41.869537] EXT4-fs (sda3): Mount option "noacl" will be removed by 3.5 Both nopti and HT disabled produced normal functionality with Buster's 4.19.0-2-i686-pae (4.19.16-1) on host gx280. Sessions were about 6 minutes long including startx/TDE and opening Firefox. I saved the Debian sessions' screens via "script" command to files in case you want more detail. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c34 --- Comment #34 from Felix Miata --- (In reply to Andreas Stenglein from comment #33)

With "graphical boot" I mean the boot menu is in high-resolution. You can switch to textmode in Yast -> system -> bootloader -> tab: kernel parameter -> box: [] Use graphical console / Graphik-Konsole benutzen UnSelecting the bootloader "graphical console" is a sufficient workaround on my side.

I'm still not sure what you want from me here. Ever since Grub as an option was terminated, the only time I see a YaST Bootloader screen on my own hardware is during installation, where, except on UEFI systems, I customarily select to install no bootloader. With the exception of UEFI, all Gnu Linux operating systems here are booted using an openSUSE Grub 0.97 version, usually 0.97.194.

To check if PTI is enabled on your fedora installation you have to look for a line Kernel/User page tables isolation: enabled in dmesg output.

dmesg | grep -i 'page table' and dmesg | grep -i isola both return null with 4.20.6-100.fc28.i686

According your description your CPU supports 64bit, too. Did you try a 64bit operating system?

I don't know how to answer this. I have hundreds of bootable operating system installations. Whether half or more are using 64bit kernels I have no idea. I didn't do my first 64bit installation on any of my own PCs until around the time openSUSE announced there would be no 32bit Leap. I think I have at most 4 PCs with both 32bit and 64bit installed, more than I care to have already, so I intend to create no more purposely. Comment #29 & #32 host gx280 was selected for the test that resulted in comment #29 because it has the fastest CPU clock on hardware on which I have running 32bit versions of openSUSE and Fedora and Debian, along with a gfxchip supported by a KMS Xorg driver, not because of its CPU's optimal arch. None of my 64bit installations exhibit this bug. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c36 --- Comment #36 from Andreas Stenglein --- (In reply to Felix Miata from comment #34)

(In reply to Andreas Stenglein from comment #33)

...

With "graphical boot" I mean the boot menu is in high-resolution. You can switch to textmode in Yast -> system -> bootloader -> tab: kernel parameter -> box: [] Use graphical console / Graphik-Konsole benutzen UnSelecting the bootloader "graphical console" is a sufficient workaround on my side.

I'm still not sure what you want from me here. Ever since Grub as an option was terminated, the only time I see a YaST Bootloader screen on my own hardware is during installation, where, except on UEFI systems, I customarily select to install no bootloader. With the exception of UEFI, all Gnu Linux operating systems here are booted using an openSUSE Grub 0.97 version, usually 0.97.194.

I don't mean some "YaST Bootloader screen" that shows up during boot, but the configuration of the bootloader inside YaST: You have to start the OS you usually configure the bootloader with, and start YaST using the menu, or by typing /usr/bin/xdg-su -c /sbin/yast2 into an xterm, then find the setting as mentioned above.

...

To check if PTI is enabled on your fedora installation you have to look for a line Kernel/User page tables isolation: enabled in dmesg output.

dmesg | grep -i 'page table' and dmesg | grep -i isola both return null with 4.20.6-100.fc28.i686

So PTI seems not enabled on your fedora(kernel). To check if fedora is in principle also affected from the bug you would have to enable PTI in your fedora(kernel) or use fedoras pae-kernel, if available; or compile yourself.

...

According your description your CPU supports 64bit, too. Did you try a 64bit operating system?

I don't know how to answer this. I have hundreds of bootable operating system installations. Whether half or more are using 64bit kernels I have no

Until now I had the opinion you have one system with different partitions and operating systems on them. And the bug only occurs on your machine with its special components and mine (both with Pentium 4 CPU, both with AMD Radeon graphics), and nobody else cares since this setup is so unusual nowadays...

idea. I didn't do my first 64bit installation on any of my own PCs until around the time openSUSE announced there would be no 32bit Leap. I think I have at most 4 PCs with both 32bit and 64bit installed, more than I care to have already, so I intend to create no more purposely. Comment #29 & #32 host gx280 was selected for the test that resulted in comment #29 because it has the fastest CPU clock on hardware on which I have running 32bit versions of openSUSE and Fedora and Debian, along with a gfxchip supported by a KMS Xorg driver, not because of its CPU's optimal arch. None of my 64bit installations exhibit this bug.

So in reality you have 4 different machines which all exhibit this bug: provided that a 32bit pae kernel is used, PTI is enabled, kernel-modeset is active and Hyperthreading is enabled, right? Do you have a box with a real multicore CPU, where you can switch Hyperthreading off and still have more threads in parallel: for checking if Hyperthreading is the culprit or just using a multi core system. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c38 --- Comment #38 from Felix Miata --- (In reply to Andreas Stenglein from comment #36) I think I'm lost as to why the subject of YaST Bootloader is in this discussion. I use YaST relatively little for anything, and never on MBR installations for bootloader except as it is present in in the installation system. I purposely have nothing I know of, other than Grub Legacy Gfxboot, that pertains to graphical boot installed anywhere it's not forced. I like my boots to have no resemblance to booting Windows, seeing that something seems to actually be happening via the kernel's nice legible boldish framebuffer font, even though it's usually too fleeting to read in real time. I don't compile anything, or use OBS other than for installing and running various packages others have built there. I do have installations other than Fedora and openSUSE, but except for some Mandrake and its progeny, are all Debian or Debian-based. I do have spare rv370 and rv380 ATI PCIe cards that might be pressed into service to help resolve this.

So in reality you have 4 different machines which all exhibit this bug: provided that a 32bit pae kernel is used, PTI is enabled, kernel-modeset is active and Hyperthreading is enabled, right?

I'm not sure, due to lack of inventory of what has PTI enabled. Otherwise, true.

Do you have a box with a real multicore CPU, where you can switch Hyperthreading off and still have more threads in parallel: for checking if Hyperthreading is the culprit or just using a multi core system.

I'm pretty sure every openSUSE here newer than 13.2 that's also genuinely multicore has been switched to 64 bit. Would it help if I tried Kernel:/stable/standard/i686/kernel-pae in a 13.2 or older installation, assuming it's possible? One of those I /might/ be able to locate. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c40 --- Comment #40 from Andreas Stenglein --- also affected: kernel-pae-5.0.0-4.1.g6bc6477.i686 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c41 --- Comment #41 from Andreas Stenglein --- (In reply to Felix Miata from comment #39)

I noticed commonality between comment 0 host gx151's X600 and comment 37 attachment, rv370 Radeon, so I took the X600 out of gx151 and tried to boot its pae kernel. It wasted little time filling the screen with debris and halting. Am I right, you used some other graphics card and the problem persists. Whats the type of 2nd graphics card that shows the problem, too?

-- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c43 --- Comment #43 from Felix Miata --- (In reply to Andreas Stenglein from comment #41)

Whats the type of 2nd graphics card that shows the problem, too?

Comment 0 Prescott host gx151: Normal use: # inxi -Gxx Graphics: Device-1: Advanced Micro Devices [AMD/ATI] RV370 [Radeon X600/X600 SE] driver: radeon v: kernel bus ID: 01:00.0 chip ID: 1002:5b62 ... Temporary change for purpose of comment 39: # inxi -Gxx Graphics: Device-1: Intel 82945G/GZ Integrated Graphics vendor: Dell driver: i915 v: kernel bus ID: 00:02.0 chip ID: 8086:2772 ... Comment 29 Northwood host gx280: # inxi -Gxx Graphics: Device-1: Intel 82915G/GV/910GL Integrated Graphics driver: i915 v: kernel bus ID: 00:02.0 chip ID: 8086:2582 ... -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c45 --- Comment #45 from Felix Miata --- Last couple of gx151 TW20190307 boots with PAE 4.20.13 with vga= and video= omitted from cmdline and hyperthreading disabled failed to produce this bug in about an hour's time. I then switched hyperthreading back on. That was well over an hour ago, and still this bug has remained apparently absent while digressing through Plasma trouble: https://lists.opensuse.org/opensuse-support/2019-03/msg00075.html -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c52 --- Comment #52 from Felix Miata --- (In reply to Jiri Slaby from comment #47)

(In reply to Andreas Stenglein from comment #46)

...

also affected: kernel-pae-5.2.rc3-3.1.gb4eda05.i686

Does blacklisting the radeon driver make the problem disappear too? It looks like radeon does not play with PTI well -- the reason is unknown to date.

Comment 39 suggests this applies to Intel GFX as well as Radeon. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c54 --- Comment #54 from Andreas Stenglein --- also affected: kernel-pae-5.2.rc4-2.1.gad82a9a.i686 kernel-pae-5.1.9-5.1.ge68f829.i686 Maybe this is also of interest: During start-up, before you can enter the bios configuration or grub gets loaded, there is a low resolution "intel inside Pentium 4 HT" graphic splash screen shown. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 Andreas Stenglein changed: What |Removed |Added ---------------------------------------------------------------------------- Summary|32bit x86 4.19 4.20 5.0 pae |32bit x86 4.19 4.20 5.0 5.1 |and vanilla kernels |5.2-rc pae and vanilla |randomly lockup or trigger |kernels randomly lockup or |reboot |trigger reboot -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c57 --- Comment #57 from Andreas Stenglein --- (In reply to Joerg Roedel from comment #56)

Can you please try to boot the systems with 'nohugeiomap' and report whether the problem disappears? Yes, with nohugeiomap the problem disappears on my system. I booted the system three times: no oops. I tried this with kernel 5.1.9-5.ge68f829-pae

-- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c59 --- Comment #59 from Felix Miata --- typo: s/5.1.9/5.1.7/ -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c65 --- Comment #65 from Felix Miata --- Negative: kernel-pae-5.1.11-1.1.g32b9efd.i686.rpm is no improvement on comment 0 host gx151. Some messages (not in order or complete): _raw_spin_lock preempt_schedule_common pgd_free watchdog: BUG: soft lockup CPU#1 stuck for 22s do_fast_syscall_32 sort_range EIP: multi_cpu_stop EIP: native_queued_spin_lock_slowpath rewind_stack_do_exit -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c74 --- Comment #74 from Andreas Stenglein --- Unfortunately this kernel 5.2.0-1.ga58fb14-pae doesn't fix the problem. However the workarounds nohugeiomap nopti modprobe.blacklist=radeon work with this kernel. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c77 Andreas Stenglein changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(Andreas.Stenglein | |@mailbox.org) | --- Comment #77 from Andreas Stenglein --- (In reply to Joerg Roedel from comment #75)

(In reply to Andreas Stenglein from comment #74)

...

Unfortunately this kernel 5.2.0-1.ga58fb14-pae doesn't fix the problem. However the workarounds nohugeiomap nopti modprobe.blacklist=radeon work with this kernel.

So you still see spontaneous reboots with this kernel? Or is there a different kind of problem you see with it?

With previous/vanilla pae kernels I almost ever got lockup/oops and most of the time I could see something going on on the screen for some time. Sometimes screen was black. With this kernel when the lockup occurs the screen goes black, nothing is visible. I tried about 10 times and can't remember seeing something going on on the screen after the lockup/oops. Once it was possible to switch the NumLock LED after lockup, and once both CapsLock and ScrollLock LEDs began blinking after some time; both happened with black screen. And this kernel booted twice (out of about 10 times) with radeon/drm error, but working desktop and software GL renderer. [ 16.601558] [drm:r100_ib_test [radeon]] *ERROR* radeon: ib test failed (scratch(0x15E8)=0xCAFEDEAD) [ 16.610832] [drm:radeon_ib_ring_tests [radeon]] *ERROR* radeon: failed testing IB on GFX ring (-22). [ 16.620090] [drm:radeon_device_init [radeon]] *ERROR* ib ring test failed (-22). I'm going to attach full dmesg output. I have to check if this kernel reboots automatically after some time. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c79 --- Comment #79 from Andreas Stenglein --- Created attachment 810378 --> http://bugzilla.opensuse.org/attachment.cgi?id=810378&action=edit dmesg of patched kernel 5.2.0-1.ga58fb14-pae with drm/radeon error, occurence #2 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c81 --- Comment #81 from Andreas Stenglein --- Created attachment 810379 --> http://bugzilla.opensuse.org/attachment.cgi?id=810379&action=edit dmesg of patched kernel 5.2.0-1.ga58fb14-pae with radeon and gart problem -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c83 --- Comment #83 from Andreas Stenglein --- Created attachment 810381 --> http://bugzilla.opensuse.org/attachment.cgi?id=810381&action=edit dmesg of patched kernel 5.2.0-1.ga58fb14-pae with ERROR but somewhat working machine -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1118689 http://bugzilla.opensuse.org/show_bug.cgi?id=1118689#c88 --- Comment #88 from Andreas Stenglein --- (In reply to Joerg Roedel from comment #87)

Upstream review found another bug in the patch. I fixed it and re-built the kernel-packages. Can you please try the latest kernel-pae from

Hello Jörg, I booted kernel 5.2.1-1.g121429f-pae 5 times without finding oops in dmesg. Then I did zypper dup, without issues. Booted another 5 times without noticing issues. It looks like you fixed the bug for me, so from my side you can close the bug when the patch is in the opensuse kernel. Thanks a lot! If I still should try with the nopti option, let me know. Otherwise I think this test isn't necessary any more. -- You are receiving this mail because: You are on the CC list for the bug.