[U-Boot-Users] minor changes brings system to hang

Hello list,
I have a strange behaviour when I try to modify my u-boot image (u-boot-1.1.0). The same results with u-boot-1.1.1. I use a ARM9 of NetSilicon, uclibc toolchain.
When I modify e.g. CONFIG_BOOTCOMMAND or add a define in my config file u-boot hangs booting. Maybe I'm blind and don't see what I have forgotten...
I disassembled the working and a not working version and compared them. They looked completely different - about 800 differences - although I deleted only two characters in the CONFIG_BOOTCOMMAND string.
What catched my eye were the start, bss_start and bss_end addresses. Here the output of the working image:
00f80044 <_armboot_start>: f80044: 00f80000 rsceqs r0, r8, r0
00f80048 <_bss_start>: f80048: 00f9e308 rsceqs lr, r9, r8, lsl #6
00f8004c <_bss_end>: f8004c: 00fa2784 rsceqs r2, sl, r4, lsl #15
and the output of the damaged one 00f80044 <_armboot_start>: f80044: 00f80000 rsceqs r0, r8, r0
00f80048 <_bss_start>: f80048: 00f9e300 rsceqs lr, r9, r0, lsl #6
00f8004c <_bss_end>: f8004c: 00fa277c rsceqs r2, sl, ip, ror r7
The output fragment I get when u-boot boots is
ot code: 00F80000 -> 00F9E308 BSS: -> 00FA2784 Stack: 0badc0de Stack: 0badc0de Configuration: #0: 00000000 16 MB?
The bss_start and bss_end addresses do not correspond to the addresses of the disassembled image.
Any suggestions?
Thanks for your help,
-- Jonas Dietsche

I use a ARM9 of NetSilicon, uclibc toolchain.
Ah, have you been able to use a uclibc toolchain to compile u-boot? I tried (briefly) with a 3.4 based toolchain (uclibc) and it fails at the end. But maybe you're not using a 3.4 compiler?
I currently use the ELDK compiler to compile u-boot and the uclibc toolchain to compile the rest (kernel, rootfs and userspace utils).
participants (2)
-
Jonas Dietsche
-
Marc Leeman