[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Jambor changed: What|Removed |Added Status|REOPENED|RESOLVED Resolution|--- |FIXED --- Comment #23 from Martin Jambor --- (In reply to Martin Liška from comment #22) > Should be fixed now. So let's assume it is.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #22 from Martin Liška --- Should be fixed now.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #21 from Martin Liška --- Author: marxin Date: Tue Jul 30 14:52:29 2019 New Revision: 273910 URL: https://gcc.gnu.org/viewcvs?rev=273910=gcc=rev Log: Use edge->indirect_unknown_callee in cgraph_edge::make_direct (PR ipa/89330). 2019-07-30 Martin Liska PR ipa/89330 * cgraph.c (cgraph_edge::make_direct): Use edge->indirect_unknown_callee as edge->resolve_speculation can deallocate edge which is this pointer. Modified: trunk/gcc/ChangeLog trunk/gcc/cgraph.c
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330
Martin Liška changed:
What|Removed |Added
Assignee|marxin at gcc dot gnu.org |jamborm at gcc dot
gnu.org
--- Comment #20 from Martin Liška ---
Ok, one can reproduce the problem with:
$ ../configure --enable-languages=c,c++,lto,fortran --disable-multilib
--prefix=/home/marxin/bin/gcc2 --enable-checking=release --without-isl
--disable-libsanitizer --disable-bootstrap
Reduced test-case:
$ cat /tmp/tree-ssa-sccvn-ice.ii
class A {
public:
int dest;
};
class B {
public:
B(int);
virtual int m_fn1();
};
int B::m_fn1() { return __null; }
void fn1(B , bool, bool, bool, bool, bool) {
for (;;) {
p1.m_fn1();
p1.m_fn1();
}
}
A a;
void fn2(bool p1) {
B b(a.dest);
fn1(b, false, false, p1, fn2, );
}
I also have a debugging patch that shows which released edge is used:
diff --git a/gcc/cgraph.c b/gcc/cgraph.c
index 81250acb70c..c585713b23a 100644
--- a/gcc/cgraph.c
+++ b/gcc/cgraph.c
@@ -1008,8 +1008,10 @@ symbol_table::free_edge (cgraph_edge *e)
if (e->m_summary_id != -1)
edge_released_summary_ids.safe_push (e->m_summary_id);
+ fprintf (stderr, "releasing: %p: %p->%p\n", e, e->caller, e->callee);
if (e->indirect_info)
ggc_free (e->indirect_info);
+ e->caller = NULL;
ggc_free (e);
}
@@ -,6 +1113,8 @@ cgraph_edge::speculative_call_info (cgraph_edge *,
if (e2->call_stmt)
{
e = e->caller->get_edge (e2->call_stmt);
+ fprintf (stderr, "cgraph_edge::speculative_call_info: %p\n", e);
+ gcc_assert (e->caller);
gcc_assert (e->speculative && !e->indirect_unknown_callee);
}
else
@@ -1223,6 +1227,8 @@ cgraph_edge::make_direct (cgraph_node *callee)
ggc_free (indirect_info);
indirect_info = NULL;
+ fprintf (stderr, "cgraph_edge::make_direct: %p\n", this);
+
/* Get the edge out of the indirect edge list. */
if (prev_callee)
prev_callee->next_callee = next_callee;
$ ./xgcc -B. /tmp/tree-ssa-sccvn-ice.ii -c -O2
/tmp/tree-ssa-sccvn-ice.ii: In member function ‘virtual int B::m_fn1()’:
/tmp/tree-ssa-sccvn-ice.ii:10:25: warning: converting to non-pointer type ‘int’
from NULL [-Wconversion-null]
10 | int B::m_fn1() { return __null; }
| ^~
releasing: 0x7fe4853894e0: 0x7fe485380438->(nil)
releasing: 0x7fe485389478: 0x7fe485380438->(nil)
releasing: 0x7fe485389820: 0x7fe4853805a0->0x7fe485380438
releasing: 0x7fe4853897b8: 0x7fe4853805a0->0x7fe485380708
releasing: 0x7fe485389958: 0x7fe485380438->(nil)
releasing: 0x7fe4853898f0: 0x7fe485380438->(nil)
releasing: 0x7fe485389af8: 0x7fe485380438->(nil)
releasing: 0x7fe485389a90: 0x7fe485380438->(nil)
releasing: 0x7fe485389a28: 0x7fe4853805a0->0x7fe485380438
releasing: 0x7fe4853899c0: 0x7fe4853805a0->0x7fe485380708
releasing: 0x7fe485389ea0: 0x7fe4853805a0->0x7fe485380438
releasing: 0x7fe485389e38: 0x7fe4853805a0->0x7fe485380708
cgraph_edge::speculative_call_info: 0x7fe485389888
cgraph_edge::speculative_call_info: 0x7fe485389888
releasing: 0x7fe4853899c0: 0x7fe4854f92d0->(nil)
cgraph_edge::make_direct: 0x7fe4853899c0
during IPA pass: inline
/tmp/tree-ssa-sccvn-ice.ii: At global scope:
/tmp/tree-ssa-sccvn-ice.ii:21:1: internal compiler error: Segmentation fault
21 | }
| ^
0x11f0c8a crash_signal
../../gcc/toplev.c:326
0x7fe48560de4f ???
/usr/src/debug/glibc-2.29-7.3.x86_64/signal/../sysdeps/unix/sysv/linux/x86_64/sigaction.c:0
0xc078da cgraph_edge::make_direct(cgraph_node*)
../../gcc/cgraph.c:1238
0xef8525 ipa_make_edge_direct_to_target(cgraph_edge*, tree_node*, bool)
../../gcc/ipa-prop.c:2978
0xef951c try_make_edge_direct_virtual_call
../../gcc/ipa-prop.c:3398
0xef97cc update_indirect_edges_after_inlining
../../gcc/ipa-prop.c:3463
0xef9b71 propagate_info_to_inlined_callees
../../gcc/ipa-prop.c:3556
0xefa1ed ipa_propagate_indirect_call_infos(cgraph_edge*, vec*)
../../gcc/ipa-prop.c:3713
0x1eb1b06 inline_call(cgraph_edge*, bool, vec*,
int*, bool, bool*)
../../gcc/ipa-inline-transform.c:486
0x1ea3efd inline_small_functions
../../gcc/ipa-inline.c:2088
0x1ea58c3 ipa_inline
../../gcc/ipa-inline.c:2550
0x1ea66b0 execute
../../gcc/ipa-inline.c:2958
@Martin: Can you please take a look at that?
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #19 from Martin Liška --- (In reply to Andreas Schwab from comment #18) > This breaks bootstrap with --enable-checking=release. > > during IPA pass: cp > ../../gcc/gcc/tree-ssa-sccvn.c:7307:1: internal compiler error: in > speculative_call_info, at cgraph.c:1114 > 7307 | } > | ^ > 0xc0dc0e cgraph_edge::speculative_call_info(cgraph_edge*&, cgraph_edge*&, > ipa_ref*&) > ../../gcc/gcc/cgraph.c:1114 > 0xc0ddbb cgraph_edge::resolve_speculation(tree_node*) > ../../gcc/gcc/cgraph.c:1155 > 0xc0e025 cgraph_edge::make_direct(cgraph_node*) > ../../gcc/gcc/cgraph.c:1215 > 0xefefaa ipa_make_edge_direct_to_target(cgraph_edge*, tree_node*, bool) > ../../gcc/gcc/ipa-prop.c:2978 > 0x1eb40da ipcp_discover_new_direct_edges > ../../gcc/gcc/ipa-cp.c:3348 > 0x1eb55b9 create_specialized_node > ../../gcc/gcc/ipa-cp.c:3913 > 0x1eb731f decide_whether_version_node > ../../gcc/gcc/ipa-cp.c:4803 > 0x1eb77c7 ipcp_decision_stage > ../../gcc/gcc/ipa-cp.c:4915 > 0x1eb7fd6 ipcp_driver > ../../gcc/gcc/ipa-cp.c:5092 > 0x1eb81b8 execute > ../../gcc/gcc/ipa-cp.c:5183 Confirmed, I've got a patch candidate that I'll send soon.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Andreas Schwab changed: What|Removed |Added Status|RESOLVED|REOPENED Resolution|FIXED |--- --- Comment #18 from Andreas Schwab --- This breaks bootstrap with --enable-checking=release. during IPA pass: cp ../../gcc/gcc/tree-ssa-sccvn.c:7307:1: internal compiler error: in speculative_call_info, at cgraph.c:1114 7307 | } | ^ 0xc0dc0e cgraph_edge::speculative_call_info(cgraph_edge*&, cgraph_edge*&, ipa_ref*&) ../../gcc/gcc/cgraph.c:1114 0xc0ddbb cgraph_edge::resolve_speculation(tree_node*) ../../gcc/gcc/cgraph.c:1155 0xc0e025 cgraph_edge::make_direct(cgraph_node*) ../../gcc/gcc/cgraph.c:1215 0xefefaa ipa_make_edge_direct_to_target(cgraph_edge*, tree_node*, bool) ../../gcc/gcc/ipa-prop.c:2978 0x1eb40da ipcp_discover_new_direct_edges ../../gcc/gcc/ipa-cp.c:3348 0x1eb55b9 create_specialized_node ../../gcc/gcc/ipa-cp.c:3913 0x1eb731f decide_whether_version_node ../../gcc/gcc/ipa-cp.c:4803 0x1eb77c7 ipcp_decision_stage ../../gcc/gcc/ipa-cp.c:4915 0x1eb7fd6 ipcp_driver ../../gcc/gcc/ipa-cp.c:5092 0x1eb81b8 execute ../../gcc/gcc/ipa-cp.c:5183
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Liška changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution|--- |FIXED --- Comment #17 from Martin Liška --- Should be fixed now.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330
--- Comment #16 from Martin Liška ---
Author: marxin
Date: Sun Jul 28 17:10:26 2019
New Revision: 273857
URL: https://gcc.gnu.org/viewcvs?rev=273857=gcc=rev
Log:
Release cgraph_{node,edge} via ggc_free (PR ipa/89330).
2019-07-28 Martin Liska
PR ipa/89330
* cgraph.c (symbol_table::create_edge): Always allocate
a cgraph_edge.
(symbol_table::free_edge): Store summary_id to
edge_released_summary_ids if != -1;
* cgraph.h (NEXT_FREE_NODE): Remove.
(SET_NEXT_FREE_NODE): Likewise.
(NEXT_FREE_EDGE): Likewise.
(symbol_table::release_symbol): Store summary_id to
cgraph_released_summary_ids if != -1;
(symbol_table::allocate_cgraph_symbol): Always allocate
a cgraph_node.
Modified:
trunk/gcc/ChangeLog
trunk/gcc/cgraph.c
trunk/gcc/cgraph.h
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Liška changed: What|Removed |Added Assignee|jamborm at gcc dot gnu.org |marxin at gcc dot gnu.org --- Comment #15 from Martin Liška --- I'll take the issue as Martin installed the patch. I'm testing right now the patch that releases deleted edges.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #14 from Martin Jambor --- Author: jamborm Date: Fri Jul 26 08:44:51 2019 New Revision: 273825 URL: https://gcc.gnu.org/viewcvs?rev=273825=gcc=rev Log: [PR 89330] Remove non-useful speculations from new_edges 2019-07-26 Martin Jambor PR ipa/89330 * ipa-inline-transform.c (check_speculations_1): New function. (push_all_edges_in_set_to_vec): Likewise. (check_speculations): Use check_speculations_1, new parameter new_edges. (inline_call): Pass new_edges to check_speculations. * ipa-inline.c (add_new_edges_to_heap): Assert edge_callee is not NULL. (speculation_useful_p): Early return true if edge is inlined, remove later checks for inline_failed. testsuite/ * g++.dg/lto/pr89330_[01].C: New test. * g++.dg/tree-prof/devirt.C: Added -fno-profile-values to dg-options. Added: trunk/gcc/testsuite/g++.dg/lto/pr89330_0.C trunk/gcc/testsuite/g++.dg/lto/pr89330_1.C Modified: trunk/gcc/ChangeLog trunk/gcc/ipa-inline-transform.c trunk/gcc/ipa-inline.c trunk/gcc/testsuite/ChangeLog trunk/gcc/testsuite/g++.dg/tree-prof/devirt.C
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #13 from Martin Jambor --- I have proposed this patch on the mailing list: https://gcc.gnu.org/ml/gcc-patches/2019-07/msg01648.html
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Jambor changed: What|Removed |Added Attachment #45730|0 |1 is obsolete|| Attachment #46544|0 |1 is obsolete|| --- Comment #12 from Martin Jambor --- Created attachment 46584 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=46584=edit Another WIP patch Thanks for tracking that down, indeed we found that the speculation was undoing inlining decisions which is something that is generally unsupported by the inliner. This patch fixes that and it indeed survives LTO bootstrap of all languages (profiled LTO bootstrap is underway). Unfortunately, it causes the following tests to fail: FAIL: g++.dg/tree-prof/devirt.C scan-tree-dump-times tracer "folding virtual function call to virtual unsigned int mozPersonalDictionary::_ZThn16" 1 FAIL: g++.dg/tree-prof/devirt.C scan-tree-dump-times tracer "folding virtual function call to virtual unsigned int mozPersonalDictionary::AddRef" 1 So far all my attempts to quickly fix this without actually having to understand what is going on in the testcase have failed. I'm afraid I'll have to look deeper into it which will take time, so far I did not even manage to reproduce the problem manually.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #11 from Martin Liška --- (In reply to Martin Jambor from comment #10) > Created attachment 46544 [details] > WIP patch > > I have written another patch that removes the edges from the vector at > the time speculation is resolved rather than preventing creation of > the edges in the first place. > > Unfortunately, the patch still trips over the added assert when LTO > bootstrapping D. So we'll have to look into it and find out where the > edges get deleted other than in check_speculations before figuring out > whether this is the right approach or not. Life cycle for edge->callee is: Hardware watchpoint 5: *$2 Old value = New value = symbol_table::create_edge (this=0x7733a100, caller=, callee=, call_stmt=0x0, count=..., indir_unknown_callee=true) at ../../gcc/cgraph.c:867 867 edge->prev_caller = NULL; (gdb) bt #0 symbol_table::create_edge (this=0x7733a100, caller=, callee=, call_stmt=0x0, count=..., indir_unknown_callee=true) at ../../gcc/cgraph.c:867 #1 0x009d0a32 in cgraph_node::create_indirect_edge (this=, call_stmt=0x0, ecf_flags=0, count=..., compute_indirect_info=false) at ../../gcc/cgraph.c:955 #2 0x009ecf92 in cgraph_edge::clone (this= -> )>, n=, call_stmt=0x0, stmt_uid=4, num=..., den=..., update_original=true) at ../../gcc/cgraphclones.c:113 #3 0x009ee828 in cgraph_node::create_clone (this=, new_decl=, prof_count=..., update_original=true, redirect_callers=..., call_duplication_hook=true, new_inlined_to=, args_to_skip=0x0, suffix=0x0) at ../../gcc/cgraphclones.c:502 #4 0x01eb48ea in clone_inlined_nodes (e= -> )>, duplicate=true, update_original=true, overall_size=0x3233d30 <_ZL12overall_size>) at ../../gcc/ipa-inline-transform.c:221 #5 0x01eb4996 in clone_inlined_nodes (e= -> )>, duplicate=true, update_original=true, overall_size=0x3233d30 <_ZL12overall_size>) at ../../gcc/ipa-inline-transform.c:236 #6 0x01eb4996 in clone_inlined_nodes ( e= -> )>, duplicate=true, update_original=true, overall_size=0x3233d30 <_ZL12overall_size>) at ../../gcc/ipa-inline-transform.c:236 #7 0x01eb4996 in clone_inlined_nodes ( e= -> )>, duplicate=true, update_original=true, overall_size=0x3233d30 <_ZL12overall_size>) at ../../gcc/ipa-inline-transform.c:236 #8 0x01eb4996 in clone_inlined_nodes (e= -> )>, duplicate=true, update_original=true, overall_size=0x3233d30 <_ZL12overall_size>) at ../../gcc/ipa-inline-transform.c:236 #9 0x01eb4996 in clone_inlined_nodes (e= -> )>, duplicate=true, update_original=true, overall_size=0x3233d30 <_ZL12overall_size>) at ../../gcc/ipa-inline-transform.c:236 #10 0x01eb55f3 in inline_call (e= -> )>, update_original=true, Python Exception There is no member or method named m_vecpfx.: new_edges=0x7fffdad0, overall_size=0x3233d30 <_ZL12overall_size>, update_overall_summary=true, callee_removed=0x0) at ../../gcc/ipa-inline-transform.c:477 #11 0x01ea6a68 in inline_small_functions () at ../../gcc/ipa-inline.c:2085 #12 0x01ea85c0 in ipa_inline () at ../../gcc/ipa-inline.c:2547 #13 0x01ea945f in (anonymous namespace)::pass_ipa_inline::execute (this=0x32978f0) at ../../gcc/ipa-inline.c:2955 #14 0x00f0ddff in execute_one_pass (pass=) at ../../gcc/passes.c:2473 #15 0x00f0ed7a in execute_ipa_pass_list (pass=) at ../../gcc/passes.c:2913 #16 0x008cbf7a in do_whole_program_analysis () at ../../gcc/lto/lto.c:456 #17 0x008cc2a6 in lto_main () at ../../gcc/lto/lto.c:628 #18 0x01073ff8 in compile_file () at ../../gcc/toplev.c:456 #19 0x01076c7a in do_compile () at ../../gcc/toplev.c:2209 #20 0x01076f5a in toplev::main (this=0x7fffdf90, argc=1068, argv=0x3267250) at ../../gcc/toplev.c:2344 #21 0x01fd06b9 in main (argc=30, argv=0x7fffe098) at ../../gcc/main.c:39 (gdb) c Continuing. Hardware watchpoint 5: *$2 Old value = New value = cgraph_edge::set_callee (this= -> )>, n=) at ../../gcc/cgraph.h:3140 3140} (gdb) bt #0 cgraph_edge::set_callee (this= -> )>, n=) at ../../gcc/cgraph.h:3140 #1 0x009d15c4 in cgraph_edge::make_direct (this= -> )>, callee=) at ../../gcc/cgraph.c:1254 #2 0x00d3da66 in ipa_make_edge_direct_to_target (ie= -> )>, target=, speculative=false) at ../../gcc/ipa-prop.c:2977 #3 0x00d3eef6 in try_make_edge_direct_virtual_call (ie= -> )>, jfunc=0x7fffc6ab4468, ctx=...) at ../../gcc/ipa-prop.c:3397 #4 0x00d3f1af in update_indirect_edges_after_inlining (cs= -> )>, node=, Python Exception There is no member or method named m_vecpfx.: new_edges=0x7fffdad0) at ../../gcc/ipa-prop.c:3462 #5 0x00d3f553 in propagate_info_to_inlined_callees (cs= -> )>, node=, Python Exception There is no member or method named m_vecpfx.: new_edges=0x7fffdad0) at ../../gcc/ipa-prop.c:3555 #6 0x00d3f58a in
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #10 from Martin Jambor --- Created attachment 46544 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=46544=edit WIP patch I have written another patch that removes the edges from the vector at the time speculation is resolved rather than preventing creation of the edges in the first place. Unfortunately, the patch still trips over the added assert when LTO bootstrapping D. So we'll have to look into it and find out where the edges get deleted other than in check_speculations before figuring out whether this is the right approach or not.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Liška changed: What|Removed |Added Target Milestone|--- |10.0
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Liška changed: What|Removed |Added Attachment #45733|0 |1 is obsolete|| --- Comment #9 from Martin Liška --- Created attachment 45765 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=45765=edit Updated patch that recycles summary IDs (tested on x86_64-linux-gnu)
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Jambor changed: What|Removed |Added Assignee|hubicka at gcc dot gnu.org |jamborm at gcc dot gnu.org --- Comment #8 from Martin Jambor --- I have proposed the fix on the mailing list: https://gcc.gnu.org/ml/gcc-patches/2019-02/msg01225.html
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #7 from Martin Liška --- Created attachment 45733 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=45733=edit Follow up patch that releases memory Survives GCC LTO bootstrap and can build Inkscape with LTO.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #6 from Martin Jambor --- Created attachment 45730 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=45730=edit Untested fix I'm testing the attached fix.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #5 from Jan Hubicka --- > Let me see if I can add the respective usefulness test to the code > deciding to speculate. I see, it is mine, sorry for blaming you :) One alternative would be also to put the indirect part of pseculative edge to the vector and lookup the direct one if speculation survives. But checking prior creation should work too. Thanks for looking into this!
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Jambor changed: What|Removed |Added CC||jamborm at gcc dot gnu.org --- Comment #4 from Martin Jambor --- The indirect inlining thingy was indeed written by me, that is true, but that was before speculative inlining was creating and disposing of edges at difficult to predict times. What happens is that: 1. In the course of inlining an edge, we call ipa_propagate_indirect_call_infos to adjust jump functions and to add newly discovered direct edges to new_edges vector so that they can be added to the heap later. 2. The speculation code in try_make_edge_direct_virtual_call decides to create speculation, so a new speculative edge is created and added to the vector. 3. Immediately after calling ipa_propagate_indirect_call_infos, check_speculations is called, which finds the edge !speculation_useful_p and removes it. But the edge already is in the vector. Let me see if I can add the respective usefulness test to the code deciding to speculate.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Jan Hubicka changed: What|Removed |Added CC||mjambor at suse dot cz --- Comment #3 from Jan Hubicka --- We may have more instances of this - I think I got rid of most of that code but once upon of time it was common in GCC to walk removed instructions and edges knowing they are in freelist. This code is originally Martin's. I will try to take a look how complex the interaction beween creation and removal is these days. One way is to teach inliner cache to maintain the list of new edges transparently without having to have extra vector.
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 Martin Liška changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2019-02-13 Assignee|unassigned at gcc dot gnu.org |hubicka at gcc dot gnu.org Ever confirmed|0 |1
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #2 from Martin Liška --- Created attachment 45691 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=45691=edit test-case
[Bug ipa/89330] IPA inliner touches released cgraph_edges
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89330 --- Comment #1 from Martin Liška --- Created attachment 45690 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=45690=edit test-case
