Subject: Re: bin/33909: modload fails with "section .eh_frame ... overlaps section .data"
To: Matthias Drochner <M.Drochner@fz-juelich.de>
From: None <khym@azeotrope.org>
List: netbsd-bugs
Date: 07/06/2006 02:47:00
On Wed, Jul 05, 2006 at 09:32:32PM +0200, Matthias Drochner wrote:
> khym@azeotrope.org said:
> > Actually, I haven't ever used LKMs on alpha...
> 
> OK. Works for me now, just don't know whether the modload/elf patch
> is worth much.

I updated my source tree and built a new -current/alpha kernel, and
modload seems to work now.

> I don't have an amd64 bux running -current, just one running 3.0 where I
> can test -current kernels on, but no -current userland.
> So could you send me the output of "modload -d -v"?

Hmm, it seems that a -current/amd64 kernel reboots my machine early
in the autoconf process (sometime around when starting cpu1, I think,
but the messages go by too fast for me to be sure). However, with my
previous kernel from July 2:

ld -R /dev/ksyms -e nv_lkmentry -o /usr/lkm/nv -Ttext 0x0 /usr/lkm/nv.o
.text: addr = 0x0 size = 0x98b2 align = 0x40
.shstrtab: addr = 0x0 size = 0x47 align = 0x1
.symtab: addr = 0x0 size = 0x2f760 align = 0x8
.strtab: addr = 0x0 size = 0x1dac6 align = 0x1
.rodata: addr = 0x98c0 size = 0xb6b align = 0x20
.eh_frame: addr = 0xa430 size = 0x16c0 align = 0x8
.data: addr = 0x10bb00 size = 0x190 align = 0x20
.data section forced to offset 0xbb00 (was 0x10bb00)
.bss: addr = 0x10bca0 size = 0x11c0 align = 0x20
modload: reserving 53248 bytes of memory
ld -R /dev/ksyms -e nv_lkmentry -o /usr/lkm/nv -Ttext 0xffffffff80647000 -Tdata 0xffffffff80652b00 /usr/lkm/nv.o
ld: section .eh_frame [ffffffff80651430 -> ffffffff80652b37] overlaps section .data [ffffffff80652b00 -> ffffffff80652c8f]
modload: can't link `/usr/lkm/nv.o' creating `/usr/lkm/nv' bound to 0xffffffff80647000