[Bug debug/94281] [8/9/10 Regression] g++: error: hash.cpp: ‘-fcompare-debug’ failure (length) since r8-5241-g8697bf9f46f36168

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94281

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #48117|0                           |1
        is obsolete|                            |

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 48118
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=48118&action=edit
gcc10-pr94281.patch

While trying to simulate all the cases that the patch is meant to handle in the
debugger (i.e. also outer_stmt being a GIMPLE_BIND and some debug stmts before
and/or after that), I've discovered a bug in gimple_seq_last_nondebug_stmt
where it would return NULL on a gimple_seq containing the GIMPLE_BIND followed
by a DEBUG_BEGIN_STMT, while obviously it should return that GIMPLE_BIND
instead.
THis updated patch should fix that too.