https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118400
Bug ID: 118400
Summary: memory leak of irange at evaluate_properties_for_edge
(ipa-fnsummary.cc:690)
Product: gcc
Version: 15.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: ipa
Assignee: unassigned at gcc dot gnu.org
Reporter: rguenth at gcc dot gnu.org
Target Milestone: ---
valgrind shows a definitive memory leak at
==9724== 40,808 bytes in 1 blocks are definitely lost in loss record 4,675 of
4,707
==9724== at 0x484349F: operator new[](unsigned long) (in
/usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==9724== by 0x1118C91: irange::maybe_resize(int) [clone .part.28]
(value-range.h:713)
==9724== by 0x111C46C: maybe_resize (value-range.cc:1339)
==9724== by 0x111C46C: irange::operator=(irange const&)
(value-range.cc:1320)
==9724== by 0xB55DE7: init (value-range.h:900)
==9724== by 0xB55DE7: value_range::operator=(vrange const&)
(value-range.h:919)
==9724== by 0xB49D90: operator= (value-range.h:928)
==9724== by 0xB49D90: evaluate_properties_for_edge(cgraph_edge*, bool,
unsigned int*, unsigned int*, ipa_auto_call_arg_values*, bool)
(ipa-fnsummary.cc:690)
which is
avals->m_known_value_ranges[i] = vr;
where the ipa_auto_call_arg_values member auto_vec<value_range, 32>
m_known_value_ranges element ends up being re-allocated on assignment.
This shows at least when building SPEC CPU 2017 500.perlbench_r
