Re: X86 4.8 kernel boot questions

From: Fenghua Yu
Date: Tue Apr 04 2017 - 13:52:21 EST


On Tue, Apr 04, 2017 at 10:25:46AM -0700, Tracy Smith, CEO wrote:
> Hi All,
>
> The production board I'm using doesn't have a JTAG available,
> upgrading to 4.8 from 4.1,.and need to understand why Linux 4.8.3
> doesn't boot on a x86_64 corei7-64. Boots with no issues with 4.1.
>
> Hangs at the last message of u-boot, the typical "Starting kernel."
> The bootcmd and arguments are given below.
>
> I see a fault FFS and a message indicating the image didn't load after failover.
>
> 1) How can I verify if the kernel image was loaded?
>
> 2) What is this FFS fault? Where do I find this.
>
> 3) Has the bootargs or bootcmd changed between 4.1 and 4.8.3?
>
> 4) If the boot cmd/arg has changed, what should the boot cmd/arg be
> for 4.8.3 to boot on x86_64 corei7-64?
>
> Initial RAM disk at linear address 0x20000000, size 11638378 bytes
>
> Kernel command line: "BOOT_IMAGE=/imgx/bzImage LABEL=BOOT
> root=/dev/ram0 imgmnt=/media/sda2 imgdir=imgx img=image.rootfs
> rootdelay=2 slub_debug=F console=ttyS1,115200 bootcount=1
> bootcount_addr=0xa4000 acpi_enforce_resources=lax pram_size=0x800000
> pram_addr=10000000 pram_loc=ddr crashkernel=128M
> memmap=0x800000$0x10000000 "EFI table at 683392c0, mmap 68339300,
> mmap size 4c0, version 1, descr. size 0x30
>
> Starting kernel ...
>
> Timer summary in microseconds:
>
> Mark Elapsed Stage
> 0 0 reset
> 1 1 board_init_r
> 105 104 board_init_f
> 10,180,048 10,179,943 id=64
> 10,221,985 41,937 id=65
> 10,356,645 134,660 main_loop
> 12,366,521 2,009,876 usb_start
> 18,747,284 6,380,763 start_kernel
> Accumulated time:
> 10,162,689 ahci
>
>
> On a 4.1.26-yocto-standard #1 SMP it boots with no issues. Basically
> same .config used in both cases except for anything deprecated between
> 4.1 and 4.8.3.
>
> root@:~# cat /proc/cmdline
> BOOT_IMAGE=/imgy/bzImage LABEL=BOOT root=/dev/ram0 imgmnt=/media/sda2
> imgdir=imgy img=image.rootfs rootdelay=2 slub_debug=F
> console=ttyS1,115200 fault=FFS bootcount=3 bootcount_addr=0xa4000
> acpi_enforce_resources=lax pram_size=0x800000 pram_addr=10000000
> pram_loc=ddr crashkernel=128M memmap=0x800000$0x10000000
>
> root@CLX3001:~# cat /proc/consoles
> ttyS1 -W- (EC p a) 4:65
> netcon0 -W- (E )

Have you tried git bisect? You may pin point a patch that causes the
issue by git bisect.

You may also send full dmesg from serial console after hang.

Add stable kernel maintainer Greg to the list.

Thanks.

-Fenghua