https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116189
--- Comment #7 from Andrew Pinski ---
make_insn_raw: cur_uid: 1084
make_insn_raw: cur_debug_uid: 21
make_insn_raw: cur_uid: 1085
make_insn_raw: cur_debug_uid: 21
cannot propagate from insn 1022 into insn 19: would increase complexity of
pattern
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116189
--- Comment #6 from Andrew Pinski ---
The code for get_max_insn_count is the following:
```
int
get_max_insn_count (void)
{
int n = cur_insn_uid;
/* The table size must be stable across -g, to avoid codegen
differences due to debug in
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116189
--- Comment #5 from Andrew Pinski ---
Note cprop.cc also allocates the hashtable a similar way but its walk over the
hashtable is ok because it does not allocate psedu-registers based on the
order.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116189
Andrew Pinski changed:
What|Removed |Added
Assignee|unassigned at gcc dot gnu.org |pinskia at gcc dot
gnu.org
La
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116189
--- Comment #3 from Andrew Pinski ---
expr_hash_table has a different order depending on debug and non-debug ...
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116189
--- Comment #2 from Andrew Pinski ---
Hmm, GCC 13 also had the pre difference but we didn't start to compare debug
failure there either.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116189
Andrew Pinski changed:
What|Removed |Added
Target Milestone|--- |14.3
--- Comment #1 from Andrew Pinski
