Re: 64bit startup

[Luca]

Hi

Il 26/05/23 03:31, Samuel Thibault ha scritto:
> GNU Mach 1.8
> ELF section header table at ffffffff80010290
> biosmem: physical memory map:
> biosmem: 000000000000000000:00000000000009f000, available
> biosmem: 00000000000009fc00:0000000000000a0000, reserved
> biosmem: 0000000000000f0000:000000000000100000, reserved
> biosmem: 000000000000100000:0000000000dddde000, available
> biosmem: 0000000000dddde000:0000000000dde00000, reserved
> biosmem: 0000000000feffc000:0000000000ff000000, reserved
> biosmem: 0000000000fffc0000:000000000100000000, reserved
> vm_page: page table size: 908750 entries (85196k)
> vm_page: DMA: pages: 4080 (15M), free: 0 (0M)
> vm_page: DMA: min:500 low:600 high:1000
> vm_page: DMA32: pages: 61440 (240M), free: 61207 (239M)
> vm_page: DMA32: min:3072 low:3686 high:6144
> vm_page: DIRECTMAP: pages: 327680 (1280M), free: 297135 (1160M)
> vm_page: DIRECTMAP: min:16384 low:19660 high:32768
> vm_page: HIGHMEM: pages: 515550 (2013M), free: 0 (0M)
> vm_page: HIGHMEM: min:25777 low:30933 high:51555
> com 2 out of range
> RTC time is 2023-05-26 01:25:34
> module 0: ext2fs --readonly --host-priv-port=${host-port} 
> --device-master-port=${device-port} 
> --multiboot-command-line=${kernel-command-line} --exec-server-task=${exec-task} 
> -T typed ${root} $(fs-task=task-create) $(task-resume)
> module 1: exec /hurd/exec $(exec-task=task-create)
> 2 multiboot modules
> Kernel Double fault trap, eip 0x0
> kernel: Double fault (8), code=0
> Stopped at  0:
> no memory is assigned to address 00000000
>
> no memory is assigned to address 00000001
> addb    %al,0(%eax)
> 0x0(
> no memory is assigned to address 00000000
> ...)
> 0xffffffff8107a712(0,8101d463,400ee0,ffffffff8101d49c,0)
> 0xffffffff8106ff27(ffffffffde739868,ffffffff8101d463,ffffffffde736fb8,0,ffffffff
> 8104ceec)
> > > > > > user space <<<<<

I can't reproduce this exactly, but I remember seeing various issues 
with kdb and printing stacktaces on x86_64 (I've been using the qemu gdb 
server so far), which might cause the double fault, possibly hiding 
another issue.

Taking the ext2fs/exec static modules and trying to boot a ramdisk I 
have instead:

GNU Mach 1.8
biosmem: physical memory map:
biosmem: 000000000000000000:00000000000009f000, available
biosmem: 00000000000009fc00:0000000000000a0000, reserved
biosmem: 0000000000000f0000:000000000000100000, reserved
biosmem: 000000000000100000:00000000003ffe0000, available
biosmem: 00000000003ffe0000:000000000040000000, reserved
biosmem: 0000000000fffc0000:000000000100000000, reserved
biosmem: 00000000fd00000000:000000010000000000, reserved
vm_page: page table size: 262096 entries (24572k)
vm_page: DMA: pages: 4080 (15M), free: 0 (0M)
vm_page: DMA: min:500 low:600 high:1000
vm_page: DMA32: pages: 61440 (240M), free: 54554 (213M)
vm_page: DMA32: min:3072 low:3686 high:6144
vm_page: DIRECTMAP: pages: 196576 (767M), free: 179955 (702M)
vm_page: DIRECTMAP: min:9828 low:11794 high:19657
com 2 out of range
RTC time is 2023-05-26 09:31:57
module 0: initrd $(ramdisk-create)
rd0: 27541504 bytes @ffffffffc4c28000
module 1: ext2fs --multiboot-command-line=${kernel-command-line} 
--host-priv-port=${host-port} --device-master-port=${device-port} 
--exec-server-task=${exec-task} -T typed gunzip:device:rd0 
$(task-create) $(task-resume)
module 2: exec --device-master-port=${device-port} $(exec-task=task-create)
3 multiboot modules
task loaded: ext2fs --multiboot-command-line=console=com0 
TERM=mach-gnu-color init=/bin/sh --host-priv-port=1 
--device-master-port=2 --exec-server-task=3 -T typed gunzip:device:rd0
task loaded: exec --device-master-port=1

start ext2fs: Unhandled exception caught by task 
ext2fs(ffffffffbc4fecd0), terminating
Dump of i386_saved_state ffffffffbc4f3868:
RAX 0000000000000007 RBX 0000000000000006 RCX 0000000000000004 RDX 
00007ffffffffe90
RSI 00007ffffffffe90 RDI 0000000000000006 RBP 00007ffffffffe80 RSP 
00007fffffffee20
R8  0000000000000000 R9  0000000000000000 R10 00000000005090fc R11 
0000000000000246
R12 0000000000000000 R13 00007fffffffff00 R14 0000000000000000 R15 
0000000000000000
FSBASE 0000000000000000 GSBASE 0000000000000000
RIP 000000000055cd15 EFLAGS 00000202
trapno 14: Page fault, error 00000004


These are printed due to a local patch I have, but anyway the ext2fs 
task starts. I tried with a 32-bit ramdisk as I still have to assemble 
the 64-bit one, but I would have expected a message about wrong ELF type 
or similar.

Also I don't have KDB enabled, but it seems your build has (from the 
"ELF section header ..." message at the beginning).

Luca