crash kernel not booting

Pratyush Anand

2017-06-08 05:08:51 UTC

Post by Mahesh V
Hi folks,
Have mostly exhausted all my options to make my ARM crash kernel boot :(.
Nothings worked so far. Any pointers as to how should I proceed further?
Following are my .config parameters that I set.
+CONFIG_KEXEC=y
+CONFIG_CRASH_DUMP=y
+CONFIG_DEBUG_LL=y
+CONFIG_DEBUG_LL_UART_NONE=y
+CONFIG_EARLY_PRINTK=y
defaults. If I set CONFIG_ARM_PATCH_PHYS_VIRT, I get a compile error
('__pv_phys_offset' undeclared here (not in a function))
CONFIG_PHYS_OFFSET=0x41500000
# CONFIG_ARM_PATCH_PHYS_VIRT is not set
# CONFIG_AUTO_ZRELADDR is not set
I am trying to load the uncompressed vmlinux (not uImage or zImage) to
0x44800000 location in my RAM (crash kernel location)
iomem has the following entry
44800000-5fbfffff : System RAM
44800000-4ddfffff : Crash kernel
I used the following kexec command to load the kernel to that location
/aruba/bin/kexec -d -p /aruba/bin/vmlinux
--command-line="console=ttyHSL1,9600n8 rdinit=/sbin/init ubi.mtd=aos0
--atags --mem-min=0x44800000

I see mail on the list now.

As I had suggested,
- No need to use crashkernel=***@1096m in the command line of second kernel
- It seems that your earlyprintk is still not working. So, it would be good to
fix that first, so that you can know if your crash kernel itself crashed very
early.

Post by Mahesh V
The crash kernel vmlinux does load at that location. Verified using dd
command (44808000 == 1149272064)
/aruba/bin # dd if=/dev/mem skip=1149272064 bs=1 count=20 | od -x
20+0000000 457f 464c 0101 0001 0000 0000 0000 0000
0 records in
20+00000020 0002 0028
records out
0000024
However when I object dump the vmlinux, it has addresses pointing to c0000000
Is this the reason why my crash kernel never boots (because the
locations in vmlinux is c0000000 and not 0x44800000 which is my start
address)
these are the last few lines after crashing the main kernel
[ 7152.560043] Code: e3a02001 e5832000 f57ff04f e3a03000 (e5c32000)
[ 7152.634583] Loading crashdump kernel...
[ 7152.678819] image start ---- image->start 44808000
[ 7152.736082] kexec start address 44808000,
[ 7152.786098] reboot_code_buffer 0xc3300000,
[ 7152.837144] reboot_code_buffer_phys 44800000,
[ 7152.891315] kexec_mach_type 4704
[ 7152.931927] Bye!
[ 7152.953920] Inside __soft_restart !! c037fb68 4187fb68
[ 7153.016338] __soft_restart final address 44800000
code snippet from process.c in static void __soft_restart(void *addr) function
phys_reset = (phys_reset_t)(unsigned long)virt_to_phys(cpu_reset);
printk(KERN_EMERG "__soft_restart final address %p \n",
(unsigned char *)addr);
phys_reset((unsigned long)addr);
/* Should never get here. */
BUG();
After cpu_reset I do not see any prints or kernel messages. How do I
debug further ?
Any pointers would help
objdump of my vmlinux
[mv_iap43_2 linux-3.4.0 (IAP-OCTOMORE)]$
../../../../aos-cmn/tools/armv7-openwrt-linux-gnu-uClibc-0.9.33.2/bin/arm-openwrt-linux-objdump
-d vmlinux | less
vmlinux: file format elf32-littlearm
c0008000: e321f0d3 msr CPSR_c, #211 ; 0xd3
c0008004: ee109f10 mrc 15, 0, r9, cr0, cr0, {0}
c0008008: eb0daf8c bl c0373e40 <__lookup_processor_type>
c000800c: e1b0a005 movs sl, r5
c0008010: e28f302c add r3, pc, #44 ; 0x2c
c0008014: e8930110 ldm r3, {r4, r8}
c0008018: e0434004 sub r4, r3, r4
c000801c: e0888004 add r8, r8, r4
c0008020: eb000009 bl c000804c <__vet_atags>
c0008024: eb1fdff5 bl c0800000 <__init_begin>
c0008028: eb000159 bl c0008594 <__create_page_tables>
c000802c: e59fd00c ldr sp, [pc, #12] ; c0008040 <stext+0x40>
c0008030: e28fe004 add lr, pc, #4
c0008034: e1a08004 mov r8, r4
c0008038: e28af010 add pc, sl, #16
c000803c: ea0daf74 b c0373e14 <__enable_mmu>
c0008040: c080005c .word 0xc080005c
c0008044: c0008044 .word 0xc0008044
c0008048: c0000000 .word 0xc0000000
Thanks in advance
regards,
Mahesh
_______________________________________________
kexec mailing list
http://lists.infradead.org/mailman/listinfo/kexec

~Pratyush