https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61183
Jason Merrill <jason at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jason at gcc dot gnu.org
--- Comment #2 from Jason Merrill <jason at gcc dot gnu.org> ---
Created attachment 44324
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44324&action=edit
dump_ggc_loc_statistics from xmalloc_failed
It's actually dying during early inlining, at least in my GATHER_STATISTICS
build. Here's the mem-stats dump and backtrace from the point where allocation
fails.
The plurality of garbage is coming from inlining:
tree-inline.c:5052 (copy_tree_r) 377222112: 45.7%
The plurality of memory still held at the crash point is from
toplev.c:918 (realloc_for_line_map) 0: 0.0% 1346560: 0.2% 1340850216: 22.7%
268434456: 94.0% 22
But the enormous allocation that kills the compiler is for expanding the
mem_usage hash table.
