Subject: Re: sparc ELF kernels with a miniroot are broken.
To: matthew green <email@example.com>
From: Simon Burge <firstname.lastname@example.org>
Date: 06/24/1999 16:18:33
matthew green wrote:
> # Sections:
> # Idx Name Size VMA LMA File off Algn
> # 0 .text 00120468 f0004000 f0004000 00000074 2**2
> # CONTENTS, ALLOC, LOAD, READONLY, CODE
> # 1 .rodata 000199cd f0124468 f0124468 001204dc 2**3
> # CONTENTS, ALLOC, LOAD, READONLY, DATA
> # 2 .data 0001547c f014de38 f014de38 00139eb0 2**3
> # CONTENTS, ALLOC, LOAD, DATA
> # 3 .bss 00023d54 f0163300 f0163300 0014f400 2**8
> # ALLOC
> # 4 .comment 00004f10 00000000 00000000 0014f400 2**0
> # CONTENTS, READONLY
> Folks, I think we have a real winner here. To my green-horned point
> of view (WRT linker/compiler issues), it looks as though the .comment
> section points to exactly the same place as the .bss does. I'm
> guessing that the READONLY attributes takes precedence on that chunk
> of code for at least the first 0x4f10 bytes of code.
> Am I mistaken, or is this an overwrite (i.e. shouldn't file off for
> .comment really be 0x14f400 + 0x23d54, or 0x173154)?
> interesting theory. i'll only note that the above quoted text is from
> the "working kernel" set, and that they both (ie, working and non) do
> exactly the same thing here.
I would guess that since the BSS section doesn't have any contents
that the file offset wouldn't change between that segment and the