This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


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

Re: Troubling bootstrap information


Mark Mitchell <mark@codesourcery.com> writes:

> CodeSourcery's autobuilds page at:
> 
>   http://www.codesourcery.com/gcc_binary_snapshots
> 
> reveals that things are not perfect on the 3.0 branch.  Three out of
> the six platforms for which we have data did not build last night:
> 
> Here's a synopsis of each of the five platforms we have:
> 
>   i686-pc-linux-gnu:
> 
>     Builds.   No testsuite regressions.  (Yay!)
> 
>   hppa2.0w-hp-hpux11.00:
>    
>     Did not build.  Bootstrap comparison failure.
> 
>   i386-unknown-freebsd4.3:
> 
>     Builds.  No testsuite regressions.  (Yay!)
> 
>   mips-sgi-irix6.5:
> 
>     Did not build.  Reason unknown; I will check.
> 
>   powerpc-unknown-linux-gnu:
> 
>     Did not build.  __SIGNAL_FRAMESIZE__ undeclared compiling
>     unwind-dw2.c.
> 
>   sparc-sun-solaris2.8:
> 
>     Builds.  No testsuite regressions.  (Yay!)
> 
> Richard, would you take a look at the powerpc problem?

I'm not sure whether the kernel changed, or gcc changed, but i noticed
this one myself a few days ago.

__SIGNAL_FRAMESIZE is in asm/ptrace.h on 2.4.5-pre3 (dunno about other
versions).
However, it's also protected by an #ifdef __KERNEL, so we never see
it, even if this file gets included.

I added the hack to #define __KERNEL__, include asm/ptrace.h, and
#undef __KERNEL__. This works fine.

However, i'm not sure what the real fix is, besides using some other
macro or something.


-- 
"There was a power outage at a department store yesterday.
Twenty people were trapped on the escalators.
"-Steven Wright


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