[Bug target/107364] [10/11/12 Regression] ICE on Via Nehemiah with --march=native

marxin at gcc dot gnu.org gcc-bugzilla@gcc.gnu.org
Mon Oct 24 20:30:30 GMT 2022


https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107364

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hjl at gcc dot gnu.org

--- Comment #9 from Martin Liška <marxin at gcc dot gnu.org> ---
...

> That means that all CENTAUR/CYRIX/NSC were broken since gcc 10 and nobody
> noticed until me ?? That sounds very unlikely, does it not ?

Yes, you are the first one who noticed!

> 
> (I dont even know what NSC is)
> 
> I'm not sure, but the "wrong" commit seems to be the very large
> 
> https://github.com/gcc-mirror/gcc/commit/
> 1890f2f0e210ef515c39728c54151372d36dd187

Yep, I know about this commit.
I suspect g:6c35d16a3925958b3a22426de0cb8e04f654b6dd where e.g. sse detection
is unified:

-  has_sse = edx & bit_SSE;
and rather:
+      else if (has_feature (FEATURE_SSE2)

is used. And then in g:792317cc777123b9cac8fc9a70fc85b01a3d7a0f newly added
get_available_features
which does the ISA features detection is called only for AMD and Intel CPUs :/


More information about the Gcc-bugs mailing list