https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116161
--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> --- (In reply to Jakub Jelinek from comment #3) > Seems the mismatches are in some var names, like > - (const_int 64 [0x40])) [ MEM <vector(2) long unsigned > int> [(short unsigned int * > *)vectp_intersect__M_impl_D185643__M_start.95744_538 + 64B]+0 S16 A64]) > - (reg:V8HI 21 xmm1 [orig:482 MEM <vector(2) long unsigned int> > [(short unsigned int * *)vectp_intersect__M_impl_D185643__M_start.95744_538 > + 48B] ] [482]) > + (const_int 64 [0x40])) [ MEM <vector(2) long unsigned > int> [(short unsigned int * > *)vectp_intersect__M_impl_D185616__M_start.95744_538 + 64B]+0 S16 A64]) > + (reg:V8HI 21 xmm1 [orig:482 MEM <vector(2) long unsigned int> > [(short unsigned int * *)vectp_intersect__M_impl_D185616__M_start.95744_538 > + 48B] ] [482]) > Haven't tracked what pass created those first, but we certainly shouldn't be > adding DECL_UIDs to DECL_NAMEs. it's fine if those do not leak into the symbol table or debug info It might confuse -fcompare-debug as that compares dumps of course.
