[9fans] 8l vs. acid?

[9fans] 8l vs. acid?

[Tim Newsham]

Hi,
    I'm seeing some strange behavior in the linking and debugging
of a small program of mine:
    http://lava.net/~newsham/quest/

When I build it and inspect it in acid, it shows the map as:
   {{"text", 0x80100020, 0x80100041, 0x00000020},
    {"data", 0x80100041, 0x80100049, 0x00000041}}

and indeed I see the data starting at 0x80100041.  However
the relocation values in the binary and the symbol table
both think that the "hello" variable starts at 0x80101000.

Now if I change the -T0x80100020 to -T0x100020 the map is
even stranger:
   {{"text", 0x00001020, 0x00001041, 0x00000020},
    {"data", 0x00002000, 0x00002008, 0x00000041}}

however, the symbols say that the _main and hello are at
0x100020 and 0x101000 respectively.

What gives?  Who's to blame here, libmach?

Tim Newsham
http://www.lava.net/~newsham/

Re: [9fans] 8l vs. acid?

[Tim Newsham]

>   I'm seeing some strange behavior in the linking and debugging
> of a small program of mine:

Some more data points:

   - using -R1 while linking fixes the issue.
   - I'm using the same flags as the pc kernels.
   - I also see weird behavior when looking at pc kernels:

     screen.c:334: int hwaccel = 1;

     $ acid 9pcdisk
     acid: *hwaccel
     0x000a7373

Is acid (libmach?) out of sync with how the kernel is
actually loaded?

Tim Newsham
http://www.lava.net/~newsham/

Re: [9fans] 8l vs. acid?

[jmk]

There are fixes on sources for this.
It's caused by there not really being any difference in the
a.out layout between the bootable kernel image and a plain
executable.

I'd not noticed that before, but to tell the truth I can't
remember ever wanting to look at the data part of a boot image
file (as opposed to the data part of a loaded boot image).

Thanks.

--jim

On Wed Jun 22 16:25:54 EDT 2005, newsham@lava.net wrote:
> Hi,
>     I'm seeing some strange behavior in the linking and debugging
> of a small program of mine:
>     http://lava.net/~newsham/quest/
> 
> When I build it and inspect it in acid, it shows the map as:
>    {{"text", 0x80100020, 0x80100041, 0x00000020},
>     {"data", 0x80100041, 0x80100049, 0x00000041}}
> 
> and indeed I see the data starting at 0x80100041.  However
> the relocation values in the binary and the symbol table
> both think that the "hello" variable starts at 0x80101000.
> 
> Now if I change the -T0x80100020 to -T0x100020 the map is
> even stranger:
>    {{"text", 0x00001020, 0x00001041, 0x00000020},
>     {"data", 0x00002000, 0x00002008, 0x00000041}}
> 
> however, the symbols say that the _main and hello are at
> 0x100020 and 0x101000 respectively.
> 
> What gives?  Who's to blame here, libmach?
> 
> Tim Newsham
> http://www.lava.net/~newsham/