Re: [Patches] ARM hard-float ABI - runtime linker checks

["Joseph S. Myers" <joseph@xxxxxxxxxxxxxxxx>]

On Mon, 24 Oct 2011, Steve McIntyre wrote:

> >(Yes, when I implemented the GDB support for the hard-float ABI I made GDB 
> >check the attributes section of an executable to determine the appropriate 
> >ABI for it.  The section view of an executable or shared library, 
> >including its attributes, is certainly of use to various tools such as 
> >debuggers, but it's not intended for use by the dynamic linker.)
> 
> Hmmm. Out of curiosity, why not? I can understand the potential for a
> performance impact, but that doesn't show up in my testing.

I don't know the specific basis of Paul's statement in 
<http://sourceware.org/ml/binutils/2007-08/msg00384.html> about the intent 
of the ABI, but it's a general part of the design of ELF that things 
relevant at runtime come from the segment rather than the section view.  
As another example, ARM unwind info is accessed via PT_ARM_EXIDX rather 
than via sections - and certainly .debug_frame isn't used at runtime for 
exception handling (although it is possible to write code living in a 
process to examine .debug_frame for the executable and shared libraries, 
just as GDB uses it from outside the process).  So if attributes were to 
be used at runtime, HJ's approach of making them mapped and accessed via 
program headers would have been the way to go.

-- 
Joseph S. Myers
joseph@xxxxxxxxxxxxxxxx
_______________________________________________
Patches mailing list
Patches@xxxxxxxxxx
http://eglibc.org/cgi-bin/mailman/listinfo/patches