[SOLVED] Slackware workstation intermittently crashing

bassmadrigal · 12-20-2017, 11:50 AM

Quote:

Originally Posted by Drakeo

I have read this thread all the way through. Comes down to it. Seems to me the issue is the 15 perecent. Bad controler on a motherboard. Warms up then crashes.

There are many reports of this freeze happening with Ryzen systems, usually when the system is idle, not when it's working hard. I am on my second motherboard with the same thing happening. Making this change to the kernel and passing the kernel option has solved my issue. I was seeing freezes every 3-5 days before this change and now I have over a month of uptime. I don't think it is related to the motherboard.

Quote:

Originally Posted by mfoley

Thanks for that clarification. I don't need to "personalize" my kernel. I was worried that the version number or something was needed for getting modules in the correct /lib/modules subdirectory. However, I had checked the "Automatically append version information to the version string" option, so I guess I'll have something there.

By default, the kernel will install the modules into /lib/modules/$KERNELVERSION/. If you end up configuring two different kernels using the same version, you can use that option to add something to the end of the version so it can be separate from the other one. Something like /lib/modules/$KERNELVERSION-$CUSTOM

Quote:

Originally Posted by mfoley

Actually, I ran this overnight and did not use the -j12 option. Again, thanks for the clarification. I was under the impression that the -j option was to specify how many cores the target kernel would use when actually running the built system. I did not realize it was for just doing the compile (make menuconfig --help doesn't clarify this either). It's just as well I didn't use the -j option anyway. The whole RYZEN issue has to do with not being able to handle strenuously utilized multi-threading. Better to single-thread the compile than find the computer had shut down in the morning!

Yeah, the -j option will just help speed up compilation allowing make to spawn more than one instance. If I build a kernel with a single job (which is the default), I think it takes about 45 minutes, where if I do it with -j16, it takes about 4-5 minutes. But there is no change in the resulting program whether you use multiple jobs or not.

And the strenuous bug Ryzen systems have had is only processors made before a certain date (I want to say some time in February, but I don't remember and I'm too lazy to check) and if your original was affected, the replacement from AMD should not be. The freezes I'm experiencing and that this fix was generated around was not tied to any strenuous work on the system. I've compiled qt5 with everything maxed out without any issue. And the main thing most would experience was a segfault during the compilation... most of the time, it doesn't affect anything other than compilation.

Aeterna · 12-20-2017, 11:53 AM

sounds like painful birth. Hopefully will work.
Few notes though:

make clean && make mrproper
should be run on sources previously used, no point of running on freshly extracted sources

make oldconfig
just copies config from /boot so if you want to modify kernel config you still need to edit config file.
if you want to get rid of excess luggage run (remove hardware options you do not have) run:

Quote:

make localyesconfig

this will actually pick up all existing hardware and running modules. Once generated edit .config for net options (firewall) and extra modules not loaded but you would like to keep

Still if you want to run kernel without initrd, make sure that used fs and disk controllers are built-in not set as modules.

-jx parameter should be -jx+1 so with 10 cores you would run -j11

does not matter if you run su or kdesu, kernel is OS agnostic (I don't even have default Slackware kernel installed)

on my 3yrs old machine kernel is compiled in 7 minutes so you should get your kernel in no time with latest cpu

good luck

Drakeo · 12-20-2017, 12:13 PM

Quote:

bassmadrigal

Lets hope the patched kernel source in current uploaded today addresses some of these issues.

Code:

 Wed Dec 20 03:05:58 UTC 2017
       testing/packages/linux-4.14.7/kernel-generic-4.14.7-x86_64-1.txz: Upgraded.
       testing/packages/linux-4.14.7/kernel-headers-4.14.7-x86-1.txz: Upgraded.
       testing/packages/linux-4.14.7/kernel-huge-4.14.7-x86_64-1.txz: Upgraded.
       testing/packages/linux-4.14.7/kernel-modules-4.14.7-x86_64-1.txz: Upgraded.
       testing/packages/linux-4.14.7/kernel-source-4.14.7-noarch-1.txz: Upgraded.
       These sources have been patched with two patches that have been submitted
       upstream but have yet to appear in the mainline or stable kernels. With the
       patches applied this kernel seems stable now on both 32 and 64-bit x86.
       Thanks to Michele Ballabio for reporting the issue to the upstream kernel
       developers, and to Ming Lei for the fix. Once these patches appear in the
       4.14.x kernel series (and barring any other major regressions), we'll be
       moving these kernels back into the main tree.

bassmadrigal · 12-20-2017, 12:28 PM

Quote:

Originally Posted by Drakeo

Lets hope the patched kernel source in current uploaded today addresses some of these issues.

Code:

 Wed Dec 20 03:05:58 UTC 2017
       testing/packages/linux-4.14.7/kernel-generic-4.14.7-x86_64-1.txz: Upgraded.
       testing/packages/linux-4.14.7/kernel-headers-4.14.7-x86-1.txz: Upgraded.
       testing/packages/linux-4.14.7/kernel-huge-4.14.7-x86_64-1.txz: Upgraded.
       testing/packages/linux-4.14.7/kernel-modules-4.14.7-x86_64-1.txz: Upgraded.
       testing/packages/linux-4.14.7/kernel-source-4.14.7-noarch-1.txz: Upgraded.
       These sources have been patched with two patches that have been submitted
       upstream but have yet to appear in the mainline or stable kernels. With the
       patches applied this kernel seems stable now on both 32 and 64-bit x86.
       Thanks to Michele Ballabio for reporting the issue to the upstream kernel
       developers, and to Ming Lei for the fix. Once these patches appear in the
       4.14.x kernel series (and barring any other major regressions), we'll be
       moving these kernels back into the main tree.

Let's hope that fixes some things, but I think the issue he's tackling is related to the reports in this thread, which deals with crashes while starting up the computer, which seem to mainly affect Intel or 32bit systems. I haven't seen anything related to Ryzen in those bug reports.

But I should probably try one of the newer 4.14 kernels without my config changes to see if the problem has been fixed... but I'm not wanting to lose this stability I have now. I should probably dig through bug reports again to see if it's still being reported on the newer 4.14 releases or on the 4.15 RCs... unfortunately, kernel.org is blocked at my work, so I can't view the bug reports.

mfoley · 12-20-2017, 01:15 PM

Quote:

Originally Posted by Aeterna

sounds like painful birth. Hopefully will work. ...
make oldconfig
just copies config from /boot so if you want to modify kernel config you still need to edit config file.

Ah! Didn't know that. So my copying of Pat's config to .config, then running 'make oldconfig' essentially had the effect of clobbering my newly copied .config, eh? Well, the diff between Pat's config and my post-make-menuconfig doesn't look too extreme, so I'll run with that and see what happens.

mfoley · 12-20-2017, 01:51 PM

OK, a hopefully minor hiccup ... I'm skipping the initrd for now and therefore bypassing drakeo's and bassmadrigal's excellent instructions on that. I did the 'make modules_install'; that seemed to work and installed the modules in /lib/modules/4.14.7. I went to the next step on the Slackware instruction which are 'cp arch/x86/boot/bzImage /boot/vmlinuz-custom-2.6.37.6'. I looked for arch/x86/boot/bzImage, but I have:

Code:

# ls -l arch/x86_64/boot/bzImage
lrwxrwxrwx 1 root root 22 2017-12-20 01:47 arch/x86_64/boot/bzImage -> ../../x86/boot/bzImage

The x86_64 bit makes sense since this is an x64 system. The problem is that this link references a non-existent file. There is no arch/x86_64/boot/bzImage. I did find vmlinux in the /usr/src/linux directory:

Code:

-rwxr-xr-x   1 root root  63480160 2017-12-20 01:46 vmlinux*

So I tried using that, but when I lilo'd, I got the message "Fatal: Kernel /boot/vmlinuz-huge-4.14.7 is too big".

So, where's my kernel? The build didn't appear to fail.

Drakeo · 12-20-2017, 02:20 PM

I just built all that with you . Last night. I grabbed Pat's package scripts edited them for the 14.14.7 and all went well.

Per my post above

bassmadrigal · 12-20-2017, 02:46 PM

Quote:

Originally Posted by mfoley

OK, a hopefully minor hiccup ... I'm skipping the initrd for now and therefore bypassing drakeo's and bassmadrigal's excellent instructions on that. I did the 'make modules_install'; that seemed to work and installed the modules in /lib/modules/4.14.7. I went to the next step on the Slackware instruction which are 'cp arch/x86/boot/bzImage /boot/vmlinuz-custom-2.6.37.6'. I looked for arch/x86/boot/bzImage, but I have:

Code:

# ls -l arch/x86_64/boot/bzImage
lrwxrwxrwx 1 root root 22 2017-12-20 01:47 arch/x86_64/boot/bzImage -> ../../x86/boot/bzImage

The x86_64 bit makes sense since this is an x64 system. The problem is that this link references a non-existent file. There is no arch/x86_64/boot/bzImage. I did find vmlinux in the /usr/src/linux directory:

Code:

-rwxr-xr-x   1 root root  63480160 2017-12-20 01:46 vmlinux*

The resulting kernel is stored in x86, but a symlink from x86_64 is created to point to the x86 version. You can use either file location to copy it over.

Quote:

Originally Posted by mfoley

So I tried using that, but when I lilo'd, I got the message "Fatal: Kernel /boot/vmlinuz-huge-4.14.7 is too big".

So, where's my kernel? The build didn't appear to fail.

You need to use the bzImage file as vmlinux is (I believe) uncompressed. According to your ls, that is 63MB, where the bzImage is probably under 10MB (can't check since I haven't built the huge kernel and only have the generic).

mfoley · 12-20-2017, 02:49 PM

Quote:

Originally Posted by Drakeo

I just built all that with you . Last night. I grabbed Pat's package scripts edited them for the 14.14.7 and all went well.

Per my post above

I've done the build in /usr/src/linux as root, so I didn't think I needed to create a package. I followed the Slackware howto that both bassmadrigal and you (as Alien-Bobs howto) referenced. I did the 'make bzImage'. I saw no error message. You did that too, right? Where did your finished kernel image end up?

Quote:

Originally Posted by Bassmadrigal

The resulting kernel is stored in x86, but a symlink from x86_64 is created to point to the x86 version. You can use either file location to copy it over.

Yes, but the x86/boot/bzImage does not exist although the symlink does! Perhaps the build failed. I'll try rebuilding after hours -- the machine is currently in use.

bassmadrigal · 12-20-2017, 03:00 PM

Quote:

Originally Posted by mfoley

Yes, but the x86/boot/bzImage does not exist although the symlink does! Perhaps the build failed. I'll try rebuilding after hours -- the machine is currently in use.

Your ls shows a symlink in the x86_64/ folder pointing to the x86/ folder. This is normal. Are you saying there's nothing in x86/? There should be no symlinks in the x86/ folder. What's the output of the following?

Code:

ls -la /usr/src/linux-4.14.7/arch/x86/boot/

It should have a lot of files in there, but one should be the bzImage.

mfoley · 12-20-2017, 05:59 PM

Quote:

Originally Posted by bassmadrigal

Your ls shows a symlink in the x86_64/ folder pointing to the x86/ folder. This is normal. Are you saying there's nothing in x86/? There should be no symlinks in the x86/ folder. What's the output of the following?

Code:

ls -la /usr/src/linux-4.14.7/arch/x86/boot/

It should have a lot of files in there, but one should be the bzImage.

here's the whole directory:

Code:

# ls -la /usr/src/linux-4.14.7/arch/x86/boot/           
total 436
drwxrwxr-x  4 root root  4096 2017-12-20 10:58 ./
drwxrwxr-x 27 root root  4096 2017-12-20 01:47 ../
-rw-r--r--  1 root root  5326 2017-12-20 01:46 .a20.o.cmd
-rw-r--r--  1 root root   896 2017-12-20 01:46 .bioscall.o.cmd
-rw-r--r--  1 root root   152 2017-12-20 01:47 .bzImage.cmd
-rw-r--r--  1 root root  5374 2017-12-20 01:46 .cmdline.o.cmd
-rw-r--r--  1 root root  1768 2017-12-20 01:46 .copy.o.cmd
-rw-r--r--  1 root root  5406 2017-12-20 01:46 .cpu.o.cmd
-rw-r--r--  1 root root  5721 2017-12-20 01:46 .cpucheck.o.cmd
-rw-r--r--  1 root root  4484 2017-12-20 01:46 .cpuflags.o.cmd
-rw-r--r--  1 root root    78 2017-12-20 01:46 .cpustr.h.cmd
-rw-r--r--  1 root root  5530 2017-12-20 01:46 .early_serial_console.o.cmd
-rw-r--r--  1 root root  5435 2017-12-20 01:46 .edd.o.cmd
-rw-rw-r--  1 root root    81 2017-12-17 09:08 .gitignore
-rw-r--r--  1 root root  4574 2017-12-20 01:47 .header.o.cmd
-rw-r--r--  1 root root  5443 2017-12-20 01:47 .main.o.cmd
-rw-r--r--  1 root root  5362 2017-12-20 01:47 .memory.o.cmd
-rw-r--r--  1 root root  1916 2017-12-20 01:46 .mkcpustr.cmd
-rw-r--r--  1 root root  5722 2017-12-20 01:47 .pm.o.cmd
-rw-r--r--  1 root root  4116 2017-12-20 01:47 .pmjump.o.cmd
-rw-r--r--  1 root root  5362 2017-12-20 01:47 .printf.o.cmd
-rw-r--r--  1 root root  5365 2017-12-20 01:47 .regs.o.cmd
-rw-r--r--  1 root root    98 2017-12-20 01:47 .setup.bin.cmd
-rw-r--r--  1 root root   669 2017-12-20 01:47 .setup.elf.cmd
-rw-r--r--  1 root root  2434 2017-12-20 01:47 .string.o.cmd
-rw-r--r--  1 root root  5326 2017-12-20 01:47 .tty.o.cmd
-rw-r--r--  1 root root  5441 2017-12-20 01:47 .version.o.cmd
-rw-r--r--  1 root root  5436 2017-12-20 01:47 .video-bios.o.cmd
-rw-r--r--  1 root root  5498 2017-12-20 01:47 .video-mode.o.cmd
-rw-r--r--  1 root root  5593 2017-12-20 01:47 .video-vesa.o.cmd
-rw-r--r--  1 root root  5424 2017-12-20 01:47 .video-vga.o.cmd
-rw-r--r--  1 root root  5438 2017-12-20 01:47 .video.o.cmd
-rw-r--r--  1 root root   135 2017-12-20 01:47 .vmlinux.bin.cmd
-rw-r--r--  1 root root   204 2017-12-20 01:46 .voffset.h.cmd
-rw-r--r--  1 root root   280 2017-12-20 01:47 .zoffset.h.cmd
-rw-rw-r--  1 root root  6767 2017-12-17 09:08 Makefile
-rw-rw-r--  1 root root  3548 2017-12-17 09:08 a20.c
-rw-rw-r--  1 root root  1961 2017-12-17 09:08 apm.c
-rw-rw-r--  1 root root  1588 2017-12-17 09:08 bioscall.S
-rw-rw-r--  1 root root  1184 2017-12-17 09:08 bitops.h
-rw-rw-r--  1 root root  8051 2017-12-17 09:08 boot.h
-rw-rw-r--  1 root root  3297 2017-12-17 09:08 cmdline.c
-rw-rw-r--  1 root root   316 2017-12-17 09:08 code16gcc.h
drwxrwxr-x  2 root root  4096 2017-12-20 10:58 compressed/
-rw-rw-r--  1 root root  1052 2017-12-17 09:08 copy.S
-rw-rw-r--  1 root root  2103 2017-12-17 09:08 cpu.c
-rw-rw-r--  1 root root  6398 2017-12-17 09:08 cpucheck.c
-rw-rw-r--  1 root root  2963 2017-12-17 09:08 cpuflags.c
-rw-rw-r--  1 root root   417 2017-12-17 09:08 cpuflags.h
-rw-rw-r--  1 root root   323 2017-12-17 09:08 ctype.h
-rw-rw-r--  1 root root  3579 2017-12-17 09:08 early_serial_console.c
-rw-rw-r--  1 root root  4223 2017-12-17 09:08 edd.c
-rw-rw-r--  1 root root 17052 2017-12-17 09:08 header.S
-rw-rw-r--  1 root root  1423 2017-12-17 09:08 install.sh
-rw-rw-r--  1 root root  4409 2017-12-17 09:08 main.c
-rw-rw-r--  1 root root  3383 2017-12-17 09:08 memory.c
-rw-rw-r--  1 root root  1383 2017-12-17 09:08 mkcpustr.c
-rw-rw-r--  1 root root   330 2017-12-17 09:08 mtools.conf.in
-rw-rw-r--  1 root root  3078 2017-12-17 09:08 pm.c
-rw-rw-r--  1 root root  1743 2017-12-17 09:08 pmjump.S
-rw-rw-r--  1 root root  6122 2017-12-17 09:08 printf.c
-rw-rw-r--  1 root root   863 2017-12-17 09:08 regs.c
-rw-rw-r--  1 root root  1102 2017-12-17 09:08 setup.ld
-rw-rw-r--  1 root root  3623 2017-12-17 09:08 string.c
-rw-rw-r--  1 root root  1086 2017-12-17 09:08 string.h
drwxrwxr-x  2 root root  4096 2017-12-20 10:58 tools/
-rw-rw-r--  1 root root  2430 2017-12-17 09:08 tty.c
-rw-rw-r--  1 root root   620 2017-12-17 09:08 version.c
-rw-rw-r--  1 root root  1956 2017-12-17 09:08 vesa.h
-rw-rw-r--  1 root root  2860 2017-12-17 09:08 video-bios.c
-rw-rw-r--  1 root root  3904 2017-12-17 09:08 video-mode.c
-rw-rw-r--  1 root root  6853 2017-12-17 09:08 video-vesa.c
-rw-rw-r--  1 root root  5817 2017-12-17 09:08 video-vga.c
-rw-rw-r--  1 root root  7341 2017-12-17 09:08 video.c
-rw-rw-r--  1 root root  3252 2017-12-17 09:08 video.h

And compressed/ for good measure:

Code:

# ls -la /usr/src/linux-4.14.7/arch/x86/boot/compressed/
total 488
drwxrwxr-x 2 root root  4096 2017-12-20 10:58 ./
drwxrwxr-x 4 root root  4096 2017-12-20 10:58 ../
-rw-r--r-- 1 root root 20983 2017-12-20 01:46 .cmdline.o.cmd
-rw-r--r-- 1 root root  4615 2017-12-20 01:47 .cpuflags.o.cmd
-rw-r--r-- 1 root root 21152 2017-12-20 01:47 .early_serial_console.o.cmd
-rw-r--r-- 1 root root 43661 2017-12-20 01:47 .eboot.o.cmd
-rw-r--r-- 1 root root  3924 2017-12-20 01:47 .efi_stub_64.o.cmd
-rw-r--r-- 1 root root  3872 2017-12-20 01:47 .efi_thunk_64.o.cmd
-rw-r--r-- 1 root root 20915 2017-12-20 01:46 .error.o.cmd
-rw-rw-r-- 1 root root    66 2017-12-17 09:08 .gitignore
-rw-r--r-- 1 root root  5450 2017-12-20 01:46 .head_64.o.cmd
-rw-r--r-- 1 root root 44035 2017-12-20 01:47 .kaslr.o.cmd
-rw-r--r-- 1 root root 21414 2017-12-20 01:46 .misc.o.cmd
-rw-r--r-- 1 root root  1959 2017-12-20 01:47 .mkpiggy.cmd
-rw-r--r-- 1 root root 22761 2017-12-20 01:47 .pagetable.o.cmd
-rw-r--r-- 1 root root   204 2017-12-20 01:47 .piggy.S.cmd
-rw-r--r-- 1 root root   853 2017-12-20 01:47 .piggy.o.cmd
-rw-r--r-- 1 root root  2550 2017-12-20 01:46 .string.o.cmd
-rw-r--r-- 1 root root   113 2017-12-20 01:46 .vmlinux.bin.cmd
-rw-r--r-- 1 root root   278 2017-12-20 01:47 .vmlinux.bin.lzma.cmd
-rw-r--r-- 1 root root   639 2017-12-20 01:47 .vmlinux.cmd
-rw-r--r-- 1 root root  4122 2017-12-20 01:46 .vmlinux.lds.cmd
-rw-r--r-- 1 root root   162 2017-12-20 01:46 .vmlinux.relocs.cmd
-rw-rw-r-- 1 root root  5208 2017-12-17 09:08 Makefile
-rw-rw-r-- 1 root root   805 2017-12-17 09:08 cmdline.c
-rw-rw-r-- 1 root root   184 2017-12-17 09:08 cpuflags.c
-rw-rw-r-- 1 root root    80 2017-12-17 09:08 early_serial_console.c
-rw-r--r-- 1 root root  3552 2017-12-20 01:47 early_serial_console.o
-rw-rw-r-- 1 root root 28110 2017-12-17 09:08 eboot.c
-rw-rw-r-- 1 root root   670 2017-12-17 09:08 eboot.h
-rw-r--r-- 1 root root 16376 2017-12-20 01:47 eboot.o
-rw-rw-r-- 1 root root  2348 2017-12-17 09:08 efi_stub_32.S
-rw-rw-r-- 1 root root   124 2017-12-17 09:08 efi_stub_64.S
-rw-r--r-- 1 root root   824 2017-12-20 01:47 efi_stub_64.o
-rw-rw-r-- 1 root root  3694 2017-12-17 09:08 efi_thunk_64.S
-rw-r--r-- 1 root root  1592 2017-12-20 01:47 efi_thunk_64.o
-rw-rw-r-- 1 root root   484 2017-12-17 09:08 error.c
-rw-rw-r-- 1 root root   223 2017-12-17 09:08 error.h
-rw-rw-r-- 1 root root  6277 2017-12-17 09:08 head_32.S
-rw-rw-r-- 1 root root 12041 2017-12-17 09:08 head_64.S
-rw-rw-r-- 1 root root 21527 2017-12-17 09:08 kaslr.c
-rw-r--r-- 1 root root 14272 2017-12-20 01:47 kaslr.o
-rw-rw-r-- 1 root root 11164 2017-12-17 09:08 misc.c
-rw-rw-r-- 1 root root  2730 2017-12-17 09:08 misc.h
-rw-rw-r-- 1 root root  2080 2017-12-17 09:08 mkpiggy.c
-rw-rw-r-- 1 root root  4744 2017-12-17 09:08 pagetable.c
-rw-r--r-- 1 root root  5280 2017-12-20 01:47 pagetable.o
-rw-rw-r-- 1 root root  1692 2017-12-17 09:08 string.c
-rw-rw-r-- 1 root root  1216 2017-12-17 09:08 vmlinux.lds.S

No chance at the moment, but I'll rebuild in a couple of hours and do the 'make mrproper' per Drakeo and NOT do the 'make oldconfig' after copying Pat's config.

bassmadrigal · 12-20-2017, 08:43 PM

Yeah, based on that, it looks like the kernel didn't build. I did just verify it built on my system using the huge kernel config from Pat on the 4.14.7 sources with only the changes to the RCU subsystem listed on that website. So, I'm wondering if you accidentally left out the bzImage portion of the make bzImage modules, or maybe you didn't capitalize the "I" in bzImage (not sure if that's case sensitive) (I just checked and it will error out if you try and use lowercase "bzimage", so you wouldn't have gotten far enough to install the modules).

Drakeo · 12-20-2017, 09:10 PM

so tired of this. going to another thread the package script I pointed to was Pats many years of work . open your eyes respect the book Pat wrote it is a package script,
bzimage to vmlinuz-generic is just a name. sorry my BDay and tired of stuff. I actually bought a distro once called them asked them what a header file was.
bzimage is a compression it is vmlinuz have fun hope the MB works.

Drakeo · 12-20-2017, 09:20 PM

you think a script is not a book read it

bassmadrigal · 12-20-2017, 10:51 PM

Quote:

Originally Posted by Drakeo

so tired of this. going to another thread the package script I pointed to was Pats many years of work . open your eyes respect the book Pat wrote it is a package script,
bzimage to vmlinuz-generic is just a name. sorry my BDay and tired of stuff. I actually bought a distro once called them asked them what a header file was.
bzimage is a compression it is vmlinuz have fun hope the MB works.

Was this directed at me? He had a vmlinux file in the root of his source directory, not a vmlinuz and it was 63MB. His arch/x86/boot/ directory didn't contain bzImage (or vmlinuz), so running Pat's script would've failed.

Plus, Pat's script looks for a file in the current working directory called bzImage or starts with vmlinuz. This allows you to run the script outside of the kernel sources if you have all the needed files in the location where you run the script. If it can't find those files, it will go to the kernel source directory and grab ./arch/$KARCH/boot/bzImage, ./System.map, and ./.config and use those for packaging. Nowhere in the script does it go to the kernel source directory and look for a vmlinuz file.