Trouble with empty sections

Daniel Jacobowitz drow@false.org
Wed Nov 8 15:10:00 GMT 2006


On Tue, Nov 07, 2006 at 07:33:03AM -0800, H. J. Lu wrote:
> On Tue, Nov 07, 2006 at 09:52:11AM -0500, Daniel Jacobowitz wrote:
> > I've been updating the ARM linker tests for global changes in linker
> > output, and I stumbled across this:
> > 
> > regexp_diff match failure
> > regexp "^   ..  ..: ........     0  NOTYPE GLOBAL DEFAULT  12 _stack$"
> > line   "   14   2: 00080000     0  NOTYPE GLOBAL DEFAULT  13 _stack"
> > FAIL: Mixed ARM/Thumb dynamic application
> > 
> >   [12] .bss              NOBITS          00010368 008368 000004 00  WA 0   0  4
> >   [13] .ARM.attributes   ARM_ATTRIBUTES  00000000 008368 000010 00     0   0  1
> > 
> > It's moved to a non-allocated section.
> > 
> >     .stack         0x80000 :
> >   {
> >     _stack = .;
> >     *(.stack)
> >   }
> > 
> > Honestly, I can't follow the discussions that have been had on this
> > subject.  Maybe this was intended; if so please let me know and I'll
> > update the test.  But moving to a non-ALLOC section is decidedly odd.
> 
> There are
> 
> [13] .ARM.attributes   ARM_ATTRIBUTES  00000000 008368 000010 00     0   0  1
> 
> It isn't an SHF_ALLOC section.

Yes, that's my point.  I would naively have expected that defining a
symbol that way in a linker script would have created a symbol with a
VMA corresponding to the program, and thus implied an ALLOC section.

Anyway, if no one thinks this is a bug since, as Alan noticed, there's
no .stack on input, then I'll just fix the failure.

-- 
Daniel Jacobowitz
CodeSourcery



More information about the Binutils mailing list