[meta-freescale] building core-image-minimalfort1042d4rdb*failsbuilding hypervisor

[Kursad Oney CW]

Robert,

>
> > Robert,
> >
> > >   to make a long story (hopefully) short, i was trying to build both
> > > the 32- and 64-bit versions of t1042d4rdb, and both failed in exactly
> > > the same place, trying to build the hypervisor recipe, so i'm open to
> > > suggestions (this is on a fully-updated fedora system):
> > >
> > > [...truncated...]
> > > | /home/rpjday/oe/builds/t1042d4rdb-64b/tmp/work/ppc64e5500-poky-linux/hypervisor/git-r3/git/libos/lib/sprintf.c: In function 'sprintf':
> > > | /home/rpjday/oe/builds/t1042d4rdb-64b/tmp/work/ppc64e5500-poky-linux/hypervisor/git-r3/git/libos/lib/sprintf.c:459:6: error: specified bound 18446744073709551615 exceeds maximum object size 9223372036854775807 [-Werror=format-truncation=]
> > > |   int ret = vsnprintf(buf, ULONG_MAX, str, args);
> > > |       ^~~
> > > | cc1: all warnings being treated as errors
> >
> > I think gcc is complaining about ULONG_MAX being the size argument.
> > Since vsnprintf() returns int, it can print at most LONG_MAX characters to
> > the buffer. I'd probably just change that argument to LONG_MAX.
> >
> > I believe the format-truncation flag is a gcc 7 thing.
>
>   i suspected as much, but i'm curious as to why this error still
> exists, and wasn't flagged by regular build testing. i don't just want
> to hack the code without hearing from others as to why this issue
> exists.

To take a wild guess, I would say the yocto version you are using probably
upgraded to gcc 7 where this new flag became a default for -Werror. If you
try an older branch or just use the recipe for gcc 6, the warning might go away.

That said the DNG team at NXP would be the definite authority on this.