Re: binutils.README

[Charles Wilson <cwilson at ece dot gatech dot edu>]

Danny Smith wrote:


> One omission in the KNOWN ISSUE #0 ("variable '<var>' can't be
> auto-imported.") is that [unsigned] long long variables can't be
> auto-imported, as reported here:
> 
> http://sources.redhat.com/ml/cygwin/2001-09/msg00843.html
> 
> To see the problem, change global_int_var in dll-helpers-0.2.8/c to
> long long and look at the assembler code for printf-ing the variable.  
>  

The original reporter provided so few details in that "report" that I 
tossed the "report" in the bit bucket.  It wasn't worth my time to 
investigate it.  You have provided slightly more information, but not 
much more.  You know, there's no law against putting some of that 
assembler code in your message, rather than telling me to go 
edit/recompile some file somewhere....

Somebody (and it ain't gonna be me) needs to investigate
   a) why long long is treated as a complex data type (if that is indeed 
the case)
   b) if so, then how "this" bug is the same as the other one.  Paul 
squashed that other bug at the source -- "if a .o file contains an 
assembler instruction in which there is a nonzero addend for a 
dll-auto-imported variable, then throw an error".  Either (a) "long 
long" is a totally diferent bug, or (b) somehow Paul's check for this 
explicit error condition is being fooled.  Why? How?


--Chuck



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/