Re: [PATCH 0/2] Repost of patches for solving the build on Fedora 36 problem

[Peter Bergner <bergner@linux.ibm.com>]

On 2/3/23 1:42 AM, Richard Biener wrote:
> On Fri, Feb 3, 2023 at 6:44 AM Michael Meissner via Gcc-patches
> <gcc-patches@gcc.gnu.org> wrote:
>>
>> I'm reposting these two patches that allow GCC to build on Fedora 36 just to be
>> clear which patches I'm talking about.  The issue is that if GCC is configured
>> with long double using the IEEE 128-bit representation, it currently cannot
>> build _mulkc3 and _divkc3 in libgcc.
> 
> It's interesting that we do not see this with openSUSE where I configure with
> 
> --with-cpu=power8 --with-tune=power9 --with-long-double-format=ieee
> --with-long-double-128
> 
> note this is ppc64le, we leave ppc64 and ppc with their default.

That's strange, Bill just retested on our ppc64le openSUSE Tumbleweed system
using basically the same configure options and sees the ICE:

/home/seurer/gcc/git/gcc-trunk/libgcc/config/rs6000/_mulkc3.c: In function '__mulkc3_sw':
/home/seurer/gcc/git/gcc-trunk/libgcc/config/rs6000/_mulkc3.c:97:1: internal compiler error: in fold_stmt, at gimple-range-fold.cc:522

He did not specify --with=cpu= or --with-tune=, which means he got
power8 defaults for both of those.  It's hard for me to believe that
--with-tune=power9 could hide the issue, but we'll try that configuration
too.  Do you have any other configure options that might affect things?

Peter