https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91062
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |ASSIGNED
Last reconfirmed| |2019-07-03
CC| |hubicka at gcc dot gnu.org
Assignee|unassigned at gcc dot gnu.org |rguenth at gcc dot
gnu.org
Ever confirmed|0 |1
--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
OK, so the issue is that IPA PTA is a simple IPA pass and does
/* Build the constraints. */
FOR_EACH_DEFINED_FUNCTION (node)
{
varinfo_t vi;
/* Nodes without a body are not interesting. Especially do not
visit clones at this point for now - we get duplicate decls
there for inline clones at least. */
if (!node->has_gimple_body_p () || node->global.inlined_to)
continue;
node->get_body ();
^^^
and get_body () ends up doing
#0 ggc_collect () at /space/rguenther/src/svn/trunk2/gcc/ggc-page.c:2174
#1 0x0000000001032dbd in execute_one_ipa_transform_pass (
node=<cgraph_node * 0x7ffff687c708 "main"/3>, ipa_pass=0x31c8c50)
at /space/rguenther/src/svn/trunk2/gcc/passes.c:2232
#2 0x0000000001032e34 in execute_all_ipa_transforms ()
at /space/rguenther/src/svn/trunk2/gcc/passes.c:2250
#3 0x0000000000ae023f in cgraph_node::get_body (
this=<cgraph_node * const 0x7ffff687c708 "main"/3>)
at /space/rguenther/src/svn/trunk2/gcc/cgraph.c:3621
#4 0x00000000014b7162 in ipa_pta_execute ()
at /space/rguenther/src/svn/trunk2/gcc/tree-ssa-structalias.c:8077
thus it collects while a pass is running... omp-simd-clone.c seems to be
the only other pass using ::get_body (also a simple IPA pass).
We can pass down a flag from get_body to not call ggc_collect which I
believe is the way to go?
