Conversation
Notices
-
polprog68k (gorplop@pleroma.m68k.church)'s status on Wednesday, 25-Dec-2024 12:36:43 JST polprog68k its 4:30 in the morning and the cirrus logic arm booted the latest stable kernel (it didnt take much work). actually, it was booting at 2:30 already but i spent 2h trying to find out why defconfig does not boot. i have some clues though
#embedded #armlinux-
polprog68k (gorplop@pleroma.m68k.church)'s status on Wednesday, 25-Dec-2024 12:37:18 JST polprog68k It is arm920t with 32MB of RAM, it belongs in a museum but it's just so cute I can't resist using it for something -
polprog68k (gorplop@pleroma.m68k.church)'s status on Saturday, 28-Dec-2024 10:18:10 JST polprog68k Update on getting Linux to work. The kernel boots but there is something weird with the initrd address. The address passed by the bootloader is different than what kernel reports. Something is adding 0x03000000 to the initrd address, and then the kernel rejects it as its outside of the physical memory.
I pass the initrd address as 0x0150 0000, and the kernel reports:
INITRD: 0x04500000+0x0033b000 is not a memory region - disabling initrd
There must be some bug in arch/arm/boot/compressed/head.S. The last working kernel version on this is 2.something, and that file is common to all 32-bit ARM code, so 4 major versions of changes may have broken something for this platform. This would also explain why defconfig does not boot. So likely I will have to fix this, or use uncompressed kernel for this build (oh, it just hit me now).
Bad news - the JTAG pins on the CPU are not connected to anything.
Good news - the bootloader on this board, RedBoot, supports remote GDB protocol :3 -
polprog68k (gorplop@pleroma.m68k.church)'s status on Saturday, 28-Dec-2024 21:16:24 JST polprog68k @wolf480pl i think it would be easier in gdb, its not a lot of code and bisecting 15 years of linux changes... uhh -
Wolf480pl (wolf480pl@mstdn.io)'s status on Saturday, 28-Dec-2024 21:16:25 JST Wolf480pl @gorplop would git bisect be useful here to narrow down when it broke, or is it easier to debug it with gdb?
-
Wolf480pl (wolf480pl@mstdn.io)'s status on Saturday, 28-Dec-2024 21:16:46 JST Wolf480pl @gorplop also, new avatar?
-
polprog68k (gorplop@pleroma.m68k.church)'s status on Saturday, 28-Dec-2024 21:16:46 JST polprog68k @wolf480pl yes, new old avatar -
polprog68k (gorplop@pleroma.m68k.church)'s status on Saturday, 28-Dec-2024 21:21:30 JST polprog68k @wolf480pl yeahh, old linux is a pain to build. id rather find the offending code and fix it. or i should try booting an uncompressed kernel, because i think the bug is in the file that boots up the decompressor -
Wolf480pl (wolf480pl@mstdn.io)'s status on Saturday, 28-Dec-2024 21:21:31 JST Wolf480pl @gorplop 4 tests get you down to 1 year of linux, the next 2-3 tests narrow it down to a single "minor" (eg. 3.1.x). In theory.
I've never done that and I don't know how cumbersome it is to build and test a kernel version.
I guess getting it to build would be the difficult part :/
-
polprog68k (gorplop@pleroma.m68k.church)'s status on Monday, 30-Dec-2024 07:47:44 JST polprog68k Update: It booted an initramdisk, and then
Welcome to Buildroot ts7250 login: 8<--- cut here --- Unable to handle kernel paging request at virtual address bf180368 when execute [bf180368] *pgd=00000000:cacocry:
-
polprog68k (gorplop@pleroma.m68k.church)'s status on Monday, 30-Dec-2024 08:05:01 JST polprog68k @philpem Good news. It just booted AND did not crash, i loggied in, it is a working linux system!! -
philpem (philpem@digipres.club)'s status on Monday, 30-Dec-2024 08:05:02 JST philpem @gorplop Oh nifty. I have a couple of TS-7200 boards I keep meaning to do something with.
polprog68k likes this. -
polprog68k (gorplop@pleroma.m68k.church)'s status on Monday, 30-Dec-2024 08:05:46 JST polprog68k oh for fucks sake
# ip 8<--- cut here --- Unable to handle kernel paging request at virtual address c12d0400 when read [c12d0400] *pgd=0120041e(bad) Internal error: Oops: d [#1] ARM -
polprog68k (gorplop@pleroma.m68k.church)'s status on Monday, 30-Dec-2024 10:08:24 JST polprog68k some findings:
- Board has 32MB RAM but it's in 4 separate zones 8M each.
- I can't seem to boot the kernel from a different zone than the first one with offset around 0x218000
- Kernel text can't be too large, with -Os I managed to build one with TCP/IP and eth drivers
- I found the source of the random crashes problem -
polprog68k (gorplop@pleroma.m68k.church)'s status on Monday, 30-Dec-2024 10:13:48 JST polprog68k @philpem if you want the config and roofs.cpio, just ping me :)
-