20000131 anomolous system header warnings

Brown, Rodney rodneybrown@pmsc.com
Sun Feb 6 17:04:00 GMT 2000


Title: RE: 20000131 anomolous system header warnings





(MS-Outlook apologies).


> Zack Weinberg wrote:
> > On Fri, Feb 04, 2000 at 10:33:59PM +1100, Rodney Brown wrote:
> > > The egcs-20000131 snapshot on i586-sco-sysv5uw7.1.0 gave 
> lots of anomalous 
> > > "defined but not used" warnings on include/sys/stat.h for 
> {f,,l}stat.
> > > These are static wrapper functions on UnixWare.


> > Try this.  It fixes the problem for me with your example, 
> but may not
> > be the right answer in general - there must have been some 
> reason why
> > we made a distinction between permanent and temporary decls here.


The current_obstack == &permanent_obstack test is used a number of times in
tree.c - to set TREE_PERMANENT which is referenced by c-common.c and
dwarfout.c . Doesn't the removal of end_temporary_allocation mean that
current_obstack will remain != &permanent_obstack thereafer?


> This fixes the general-case symptoms for me, too.  Without it,
> trainloads of warnings are spewewd during a full bootstrap.







More information about the Gcc-bugs mailing list