[klibc] klibc loader crash...

[H. Peter Anvin]

Paweł Sikora wrote:
> Hi,
> 
> This is not a klibc bug. It is a kernel issue.
> 
> Fixed in 2.6.14rc4-git1: [PATCH] binfmt_elf bss padding fix
>  
>  Nir Tzachar <tzachar at cs.bgu.ac.il> points out that if an ELF file specifies a
>  zero-length bss at a whacky address, we cannot load that binary because
>  padzero() tries to zero out the end of the page at the whacky address, and
>  that may not be writeable.
>  
>  See also http://bugzilla.kernel.org/show_bug.cgi?id=5411
>  
>  So teach load_elf_binary() to skip the bss settng altogether if the elf file
>  has a zero-length bss segment.
> 

Hm.  This still confuses me.

__libc_init() should always allocate three pointer values, so .bss 
should never be empty with klibc-generated code.c.  utils/true.c looks 
like the following:

int main(void)
{
   return 0;
}

... and it generates the following headers when compiled for i386:

Sections:
Idx Name          Size      VMA       LMA       File off  Algn
   0 .text         00000070  08048094  08048094  00000094  2**2
                   CONTENTS, ALLOC, LOAD, READONLY, CODE
   1 .data         00000004  08049104  08049104  00000104  2**2
                   CONTENTS, ALLOC, LOAD, DATA
   2 .bss          0000000c  08049108  08049108  00000108  2**2
                   ALLOC

... hmmm ...

Program Headers:
   Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align
   LOAD           0x000000 0x08048000 0x08048000 0x00104 0x00104 R E 0x1000
   LOAD           0x000104 0x08049104 0x08049104 0x00004 0x00010 RW  0x1000
   GNU_STACK      0x000000 0x00000000 0x00000000 0x00000 0x00000 RWE 0x4

  Section to Segment mapping:
   Segment Sections...
    00     .text
    01     .data .bss
    02

... looks like the linker packs the .bss into the data segment if it's 
small enough.  Thus, the kernel doesn't see it.

This is a very good reason to fix the kernel.

	-hpa