https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126159

--- Comment #5 from rguenther at suse dot de <rguenther at suse dot de> ---
On Wed, 8 Jul 2026, liuhongt at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126159
> 
> --- Comment #4 from Hongtao Liu <liuhongt at gcc dot gnu.org> ---
> 
> > from dr_analyze_innermost, so the reason is you have a union here and
> > we refrain from touching those (for fear?  I don't exactly remember,
> > but most definitely if you'd had two different records in the union
> > we'd get confused big time - it's possibly fine for "scalar" elements
> > and of elements of same size).
> > 
> > Is this practically relevant?
> 
> Reduced from 708.sqlite_r : sqlite3.c:189588 / :189594  (rtree cellUnion)
> 
> union RtreeCoord {
>   RtreeValue f;      /* Floating point value */
>   int i;             /* Integer value */
>   u32 u;             /* Unsigned for byte-order conversions */
> };
> 
> NOTE: I'm just let AI go through all missed vectorization cases and try to
> reduce testcases from it, it's not necessary that the missed vectorization is
> in the the hotspot, just for vectorization capability.

OK, thanks - I'm testing the fix.

Reply via email to