This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Convert old prototype.


On Fri, 14 Jun 2013, Ondrej Bilka wrote:

> On Mon, Jun 10, 2013 at 12:28:34PM +0000, Joseph S. Myers wrote:
> > On Mon, 10 Jun 2013, Ondrej Bilka wrote:
> > 
> > > This patch fixes plural.y prototype.
> > 
> > I don't think plural.c should be regenerated with an older version of 
> > Bison (my presumption is to use the current GNU release of Bison when 
> > regenerating this file).
> > 
> > > This got lot of files, when I looked at diff it is similiar to following
> > > and I did not found nothing wrong in nptl/pthread_mutex_timedlock.c
> > > source.
> > 
> > That seems strange; it looks like it's changing function parameters.  This 
> > disassembly isn't showing relocations so we can't see what functions are 
> > being called.  Are these assertions by any chance, affected by changes to 
> > __LINE__?
> >
> Could not be this that sequence of tokens changes numbering of
> identifiers and they are changed accordingly? 

Identifier numbering (for GCC-internal versions of identifiers, e.g. for 
function-local statics) is relevant for e.g. reordering / changing address 
offsets.  It shouldn't change constant integer arguments to functions, 
which was the change there; those should be whatever integers actually 
appear in the source code rather than anything to do with 
compiler-internal numbering of identifiers that in C terms do not have 
file scope.

-- 
Joseph S. Myers
joseph@codesourcery.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]