From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andrew Wozniak <awozniak@mc.com>
Date: Sun, 17 Apr 2005 17:04:48 -0400
Subject: [U-Boot-Users] hello_world crashes U-Boot
In-Reply-To: <4262B86B.9030602@mc.com>
References: <4262B86B.9030602@mc.com>
Message-ID: <4262CF70.5020508@mc.com>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

Hi, I spent more time invstigating and found that when I use {3} as init 
data, U-Boot no longer hangs.

I also captured the hello_world .bin and .map files for the two (OK,BAD) 
cases and found these binary difference:

OK: 00000000:  00 01 82 50 94 21 FF E0 7C 08 02 A6 BF 81 00 10
BAD:00000000:  94 21 FF E0 7C 08 02 A6 BF 81 00 10 90 01 00 24

Interesting that the "entry" location has shifted.

Also, the BAD .map file is missing the following entries:

<snip>
  *(.rodata.rodata.*.gnu.linkonce.r.*)
  .rodata        0x00040260        0x8 hello_world.o

...

  *(.got2)
  .got2          0x00050268        0x4 hello_world.o
<snip>

Could all of this be a GCC bug?  I am running ppc_405-gcc version 3.2.1

Thanks again for any suggestions.


Andrew Wozniak wrote:
> Hi,
> 
> I still need to investigate this further, but figured that a post might 
> encourage some suggestions about this interesting issue.
> 
> hello_word was built with the default Makefile for the PPC440GX and it 
> runs just fine. But when it is modified so that "no" stdio is used and a 
> structure init is attempted, it crashed U-Boot (see attachment) when 
> hello_world returns.
> 
> int hello_world (int argc, char *argv[])
> {
>         struct pci_controller thehose = {0};
> 
>     app_startup(argv);
>     return (0);
> }
> 
> The interesting part is that when one or more stdio functions is added, 
> printf() for example, there is no crash.
> 
> I was hoping that someone else might be able to reproduce this crash.
> 
> My current U-Boot image has about 8 exported functions in addition to 
> the default ones. hello_world was built against this image to insure ABI 
> compatibility.
> 
> Thanks for any suggestions, Andrew
> 
> 
> ------------------------------------------------------------------------
> 
> => go 40004
> ## Starting application at 0x00040004 ...
> ## Application terminated, rc = 0x0
> NIP: 00000000 XER: 00000000 LR: 00000000 REGS: 00365c20 TRAP: 0700 DAR: 00000000
> MSR: 00029000 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 00
> 
> GPR00: 00000000 00365D10 0000FFFF 00000000 0000000A 00000001 00000001 8D3F9F08
> GPR08: 003D0900 00000001 877832C6 00000005 00000000 21000000 003F3100 00446000
> GPR16: 80050000 00028100 00000000 80000002 00009000 00365AF8 00000000 00000001
> GPR24: 00365E28 00000000 00365DF8 003C819C 0EFB0311 00000001 003D08FC 00365DF8
> ** Illegal Instruction **
> Call backtrace:
> 00000000 03000000
> Program Check Exception
> NIP: 00000000 XER: 00000000 LR: 00000000 REGS: 00365c20 TRAP: 0700 DAR: 00000000
> MSR: 00029000 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 00
> 
> GPR00: 00000000 00365D10 0000FFFF 00000000 0000000A 00000001 00000001 8D3F9F08
> GPR08: 003D0900 00000001 877832C6 00000005 00000000 21000000 003F3100 00446000
> GPR16: 80050000 00028100 00000000 80000002 00009000 00365AF8 00000000 00000001
> GPR24: 00365E28 00000000 00365DF8 003C819C 0EFB0311 00000001 003D08FC 00365DF8
> ** Illegal Instruction **