Re: Off-by-one error in ld documentation

[Nick Clifton]

Hi Hans-Peter,

> > > >    extern char start_of_ROM, end_of_ROM, start_of_FLASH;
> > > >    memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - &
> > > > start_of_ROM);


> While we're at that example, is there a reason why it is
> pretending that the symbols are for variables whose address is
> then taken (to get the proper symbol address instead of the
> faked variable values), and that the sections aren't actually
> vectors of char (unsigned if you will)?  This would IMHO yield
> more intuitive code, where newbies wouldn't be confused by
> and/or forget the address operator.  To wit:
>
> extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
> memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);

Actually that *is* a nicer way of writing it.

But, I wrote the original example based upon code that a customer had 
written (without the ampersand operators) and then complained that it 
did not work.  The point of the example therefore was to catch the 
reader's eye as code that they might write, and then show them how it 
really should be written.

I agree however that your way is better, so how does this addition to 
the linker manual sit with you ?

Cheers
  Nick

diff --git a/ld/ld.texinfo b/ld/ld.texinfo
index d20a59f..be1d490 100644
--- a/ld/ld.texinfo
+++ b/ld/ld.texinfo
@@ -3887,6 +3887,19 @@ Then the C source code to perform the copy would be:
 @end smallexample

 Note the use of the @samp{&} operators.  These are correct.
+Alternatively the symbols can be treated as the names of vectors or
+arrays and then the code will again work as expected:
+
address@hidden
address@hidden
+  extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
+
+  memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);
address@hidden group
address@hidden smallexample
+
+Note how using this method does not require the use of @samp{&}
+operators.

 @node SECTIONS
 @section SECTIONS Command