Message ID | 20140323105031.GB18530@mail.duskware.de |
---|---|
State | New |
Headers | show |
On Sun, Mar 23, 2014 at 3:50 AM, Martin Husemann <martin@duskware.de> wrote: > The flag WCONTINUED for waitpid() is not portable. > > The attached patch just defines it to 0 if missing. I don't understand why that code passes either WUNTRACED or WCONTINUED. Why not just pass 0? Ian
On Sun, Mar 23, 2014 at 03:18:42PM -0700, Ian Lance Taylor wrote: > On Sun, Mar 23, 2014 at 3:50 AM, Martin Husemann <martin@duskware.de> wrote: > > The flag WCONTINUED for waitpid() is not portable. > > > > The attached patch just defines it to 0 if missing. > > I don't understand why that code passes either WUNTRACED or > WCONTINUED. Why not just pass 0? Note the patch Martin has posted is not needed, I've committed a change to define WCONTINUED to 0 if not defined several days ago as obvious. Adding Honza in the loop why he chose to use WUNTRACED | WCONTINUED. Jakub
--- gcc/lto/lto.c.orig 2014-02-24 23:58:44.000000000 +0100 +++ gcc/lto/lto.c 2014-03-21 16:43:05.000000000 +0100 @@ -2470,6 +2470,11 @@ do_stream_out (char *temp_filename, lto_ /* Wait for forked process and signal errors. */ #ifdef HAVE_WORKING_FORK + +#ifndef WCONTINUED +#define WCONTINUED 0 +#endif + static void wait_for_child () {