Re: Bootstrap failure in stage1 fix-header, !HAVE_DESIGNATED_INITIALIZERS

[Zack Weinberg <zack at wolery dot cumb dot org>]

On Sun, Aug 20, 2000 at 01:11:47AM -0400, Kaveh R. Ghazi wrote:
>  > From: Zack Weinberg <zack@wolery.cumb.org>
>  > 
>  > On Sat, Aug 19, 2000 at 10:06:39PM -0400, Kaveh R. Ghazi wrote:
>  > > Recent CVS dies in stage1 whenever one uses cc to bootstrap on
>  > > mips-sgi-irix6.2 and sparc-sun-solaris2.7.  Fix-header gobbles up
>  > > memory until it runs out and xmalloc fails.
>  > > 
>  > > The trigger appears to be the HAVE_DESIGNATED_INITIALIZERS else-clause
>  > > since if I force the compiler to take the else path, it always fails
>  > > even with gcc.
>  > 
>  > Bizarre; none of the code under !HAVE_DESIGNATED_INITIALIZERS should
>  > be allocating memory.
>  > 
>  > I'll look into this as soon as I get a chance, but if you could change
>  > xmalloc so it aborts on failure, run the offending process under the
>  > debugger, and send me a backtrace, I'd much appreciate it.
>  > zw
> 
> Ok.
> 
> BTW, I think I found a hint as to why it dies.  The _cpp_IStable and
> _cpp_trigraph_map arrays are never initialized.

Argh.  Right.  Add a call to cpp_init() just before the call to
cpp_reader_init(&scan_in) in fix-header.c:read_scan_file.  That should
fix the problem.  (If it does, please check in a patch to that effect.)

zw