Re: XCOFF relocation data gets mangled by ld.

[Jason Sullivan]

Here's the file as it assembles with AIX 'as'.  How well does linux run on
one of those Power boxes (or is it a real PowerPC)?  Thanks!

(See attached file: assem_test4.o)
--
Jason M. Sullivan
PowerPC Embbedded Processor Solutions
IBM Microelectronics
address@hidden (919.543.5650)


                                                                                
                                                       
                      Alan Modra                                                
                                                       
                      <address@hidden         To:      Jason 
Sullivan/Raleigh/address@hidden                                               
                      net.au>                  cc:      address@hidden          
                                                 
                                               Subject: Re: XCOFF relocation 
data gets mangled by ld.                                  
                      08/02/02 02:30                                            
                                                       
                      AM                                                        
                                                       
                                                                                
                                                       
                                                                                
                                                       




On Thu, Aug 01, 2002 at 11:40:47AM -0600, Jason Sullivan wrote:
> When I send it through the AIX ld program, I get the following errors...
>
> ld: 0711-787 ERROR: Symbol .vp_branchafvt (entry 2) in object
> vp_branchafvt.o:
>         The instruction at address 0x00000020 is not a branch. It is
> 0x01204180.        RLD type: R_BA;   RLD length: 16;    Referenced
symbol:
> .vp_branchafvt

Hmm, looks like AIX ld is picking up the 2 low bytes from the previous
instruction, and the two high bytes of the branch insn.

  1c:   7c a8 01 20     mtcrf   128,r5
  20:   41 80 00 2b     bltla   28 <.vp_branchafvt+0x28>
                        20: R_BA_16     .vp_branchafvt+0x20

I'd say the R_BA_16 reloc is expected to be at offset 0x22.  Would
you mind assembling your test file under AIX and posting the resulting
object, please?  My rs6000 box has only one disk, and I blew away AIX
in favour of linux.

--
Alan Modra
IBM OzLabs - Linux Technology Centre

assem_test4.o
Description: Binary data