Re: Booting Kernel on Amiga 3000
On Tue, 6 Sep 2022, Laurent Vivier wrote:
> Le 06/09/2022 à 17:52, Andreas Schwab a écrit :
> > On Sep 06 2022, Eero Tamminen wrote:
> >
> >> Hi,
> >>
> >> On 6.9.2022 4.46, Stephen Walsh wrote:
> >>> [ 3.110000] Call Trace: [<00354d80>] panic+0xc4/0x246
> >>> [ 3.110000] [<00355744>] _printk+0x0/0x18
> >>> [ 3.110000] [<00355756>] _printk+0x12/0x18
> >>> [ 3.110000] [<001ed30c>] strlen+0x0/0x14
> >>> [ 3.110000] [<0051054a>] mount_block_root+0x17a/0x194
> >>> [ 3.110000] [<001ed268>] strncmp+0x0/0x34
> >>> [ 3.110000] [<005105ac>] mount_root+0x48/0x14e
> >>> [ 3.110000] [<005105d6>] mount_root+0x72/0x14e
> >>> [ 3.110000] [<00008001>] atari_get_hardware_list+0x6d/0x3e0
> >>
> >> This seems a bit odd line for kernel boot on Amiga.
> >
> > It's a false match (0x8001 can be anything).
> >
>
Expanding a little on what Andreas said above: Linux/m68k doesn't produce
precise backtraces. What gets printed is any value found on the stack that
could possibly be interpreted as an instruction pointer i.e. any address
within a kernel code segment.
> Stack corruption?
>
No, i's not an oops, it's normal behaviour:
[ 3.110000] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)
This expected, and it is not the crash that was originally reported.
Regarding the oops that was originally reported, the kernel already told
us why it crashed:
[ 0.000000] Ignoring memory chunk at 0x7800000:0x800000 before the first chunk
[ 0.000000] Fix your bootloader or use a memfile to make use of this area!
Hence Geert's questions aimed at figuring out what's going on with the
bootloader.
> 0x8001 looks like
>
> uapi/asm/bootinfo-amiga.h:#define BI_AMIGA_AUTOCON 0x8001 /* AutoConfig
> device */
>
By the time the kernel gets to mounting the rootfs, the stack frame
generated by amiga_parse_bootinfo() has been clobbered many times over.
Reply to: