https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117350
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|--- |15.0
Ever confirmed|0 |1
CC| |hubicka at gcc dot gnu.org
Summary|ICE in pretty print during |[15 Regression] ICE in
|bootstrap |pretty print during
| |bootstrap
Status|UNCONFIRMED |NEW
Component|ipa |gcov-profile
--- Comment #13 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Actually I am kinda of wrong about what is going on but right about why though.
So autofdo is looking up a decl but since we have free'd the front-end specific
part of the decls but for some reason the decl's decl_assembler_name was not
set during the free time we crash.
#9 0x00000000024e1007 in autofdo::string_table::get_index_by_decl
(this=0x34c0b9c0, decl=<optimized out>)
So this was exposed by r15-4610-gbf43fe6aa966ea which started to use
std::unique_ptr .
So the question is why the constructor's decl (<function_decl 0x7f17c25e7f00
__ct >) didn't gets assembler name set during pass_ipa_free_lang_data .
#8 0x00000000016ad32b in decl_assembler_name (decl=<function_decl
0x7f17c25e7f00 __ct >) at ../../gcc/gcc/tree.cc:728
