A few days ago, I tried to upgrade from 12-STABLE to 13-STABLE but the new kernel panicked instantly. Here's a transscript of the screen photo I made: avail memory = 1552841136 (14889 MB) Event timer "LAPIC" quality 600 ACPI APIC Table: <ALASKA A M I> FreeBSD/SMP: Multiprocessor System Detected: 8 CPUs FreeBSD/SMP: 1 package(s) x 4 core(s) x 2 hardware threads kernel trap 12 with interrupts disabled Fatal trap 12: page fault while in kernel mode cpuid = 0; apic id = 00 fault virtual address = 0x28 fault code = supervisor read data, page not present instruction pointer = 0x20:0xffffffff80396715 stack pointer = 0x28:0xffffffff80e17f70 frame pointer = 0x28:0xffffffff80e17f80 code segment = base rx0, limit 0xfffff, type 0x1b = DPL 0, pres 1, long 1, def32 0, gran 1 processor eflags = resume, IOPL = 0 current process = 0 () trap number = 12 panic: page fault cpuid = 0 time = 1 Uptime: 1s original photo: http://www.volki.at/austausch/panic_2022-03/screenshot_panic.jpg kernel: http://www.volki.at/austausch/panic_2022-03/kernel kernel config: http://www.volki.at/austausch/panic_2022-03/CU08 Of course, I could try a GENERIC kernel and, if it works, change line by line until the panic happens again, but this is a production system so I want to avoid experiments as long as I don't know if they are necessary.
could try enter the bootloader, then use the commands set hw.vga.acpi_ignore_no_vga="1" boot and see if its booting.
No. Panic again: http://www.volki.at/austausch/panic_2022-03/screenshot_panic1.jpg Then I updated the kernel sources (apparently to post-3.1) and replaced syscons with vt. The panic screenshot now contains a few more lines: http://www.volki.at/austausch/panic_2022-03/screenshot_panic1.jpg (this time I didn't set hw.vga.acpi_ignore_no_vga="1") It may be worth to mention that "make installkernel" finishes with an error message: ... kldxref /boot/kernel kldxref: error while reading /boot/kernel/iwlwifi-9000-pu-b0-jf-b0-46.ucode.ko: Bad address kldxref: error while reading /boot/kernel/iwlwifi-9260-th-b0-jf-b0-46.ucode.ko: Bad address -------------------------------------------------------------- >>> Installing kernel CU08 completed on Wed Mar 30 20:47:41 CEST 2022 -------------------------------------------------------------- Also suspicious: The generic kernel fails to build: # make buildkernel KERNCONF=GENERIC ... ctfconvert -L VERSION -g vers.o linking kernel.full ld: error: undefined symbol: __ashlti3 >>> referenced by sli4.c:0 (/usr/src13/sys/dev/ocs_fc/sli4.c:0) >>> sli4.o:(sli_cmd_reg_fcfi) >>> referenced by sli4.c:0 (/usr/src13/sys/dev/ocs_fc/sli4.c:0) >>> sli4.o:(sli_cmd_reg_fcfi_mrq) >>> referenced by sati_util.c:0 (/usr/src13/sys/dev/isci/scil/sati_util.c:0) >>> sati_util.o:(sati_ata_download_microcode_construct) *** Error code 1 Stop. make[2]: stopped in /usr/obj/usr/src13/amd64.amd64/sys/GENERIC *** Error code 1 *** Error code 1
Sorry, wrong links. The new panic screenshots are: http://www.volki.at/austausch/panic_2022-03/screenshot_panic2.jpg http://www.volki.at/austausch/panic_2022-03/screenshot_panic3.jpg
There is insufficient information to make progress here. At a minimum you'd need to build a kernel with KDB and KDB_TRACE options to print a backtrace upon panic. If you're no longer in a position to reproduce this please add a comment to that effect.
I managed to resolve this for me by adding the options from the MINIMAL kernel that weren't in my custum kernel. I think the culprit was I had to add: options NUMA I don't know what "Non-Uniform Memory Architecture" means and I'm not aware of having such hardware. It would be great to explain that in NOTES and mention the kernel panic at an early boot stage without that option.
(In reply to Friedrich Volkmann from comment #5) NUMA refers to SMP (multiprocessor) systems that have different access characteristics for different memory regions. Lack of that option should (of course) not result in a panic, but if the option is always included in GENERIC and MINIMAL kernel configurations without it may not be tested regularly.
My mainboard is an MSI Z97S SLI Plus with Intel Z97 Express (Haswell) chipset. My guess is that the onboard GPU makes the memory architecture non-uniform by occupying some of the system RAM.