[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

--- Comment #14 from Eric Botcazou  ---
Author: ebotcazou
Date: Mon Dec 21 15:15:50 2015
New Revision: 231882

URL: https://gcc.gnu.org/viewcvs?rev=231882&root=gcc&view=rev
Log:
PR tree-optimization/65337
* tree-ssa-pre.c (eliminate): Also clean up abnormal edges if need be.

Modified:
trunk/gcc/ChangeLog
trunk/gcc/tree-ssa-pre.c

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

--- Comment #13 from Jan Hubicka  ---
Author: hubicka
Date: Sun Dec 20 05:50:29 2015
New Revision: 231856

URL: https://gcc.gnu.org/viewcvs?rev=231856&root=gcc&view=rev
Log:
PR middle-end/65337
* tree-ssa-dce.c (bb_postorder): New static var.
(forward_edge_to_pdom): Remove.
(remove_dead_stmt): Instead of redirecting edges only keep an edge
on a path to nearest live BB.
(eliminate_unnecessary_stmts): Free bb_postorder.
* cfganal.c (dfs_find_deadend): Add START_POINTES.
* cfganal.h (inverted_post_order_compute): Update prototype.

Modified:
trunk/gcc/ChangeLog
trunk/gcc/cfganal.c
trunk/gcc/cfganal.h
trunk/gcc/tree-ssa-dce.c

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

Richard Biener  changed:

   What|Removed |Added

   Target Milestone|5.3 |5.4

--- Comment #12 from Richard Biener  ---
GCC 5.3 is being released, adjusting target milestone.

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

Richard Biener  changed:

   What|Removed |Added

   Target Milestone|5.2 |5.3

--- Comment #11 from Richard Biener  ---
GCC 5.2 is being released, adjusting target milestone to 5.3.

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[hubicka at ucw dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

--- Comment #10 from Jan Hubicka  ---
> /home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/prev-gcc/gnat1
> 
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to process 16017]
> 0x00c354ed in ei_edge (i=...) at
> /home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/basic-block.h:460
> 460   return EDGE_I (ei_container (i), i.index);
> (gdb) thread apply all bt full
> 
> Thread 2 (process 16017):
> #0  0x00c354ed in ei_edge (i=...) at
> /home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/basic-block.h:460
> No locals.
> #1  0x00c376cf in inverted_post_order_compute (post_order=0x2e3d9a0,
> start_points=0x2bb0858 ) at
> /home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/cfganal.c:837

I see, it looks like a bug in inverted_post_order_compute that is different ICE
I saw at
ppc64.  I will take a look.

Honza

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[steven at uplinklabs dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

--- Comment #9 from Steven Noonan  ---
Sure.

$ gdb --args
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/./prev-gcc/xgcc
-B/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/./prev-gcc/
-B/usr/x86_64-unknown-linux-gnu/bin/ -B/usr/x86_64-unknown-linux-gnu/bin/
-B/usr/x86_64-unknown-linux-gnu/lib/ -isystem
/usr/x86_64-unknown-linux-gnu/include -isystem
/usr/x86_64-unknown-linux-gnu/sys-include-c -g -O2 -flto=jobserver
-frandom-seed=1  -gnatpg  -W -Wall -nostdinc -I- -I. -Iada/generated -Iada
-I/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/ada
-I/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/ada/gcc-interface
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/ada/comperr.adb
-o ada/comperr.o
GNU gdb (GDB) 7.9.1
Copyright (C) 2015 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later 
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-unknown-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
.
Find the GDB manual and other documentation resources online at:
.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/./prev-gcc/xgcc...done.
warning: File
"/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/gcc/.gdbinit"
auto-loading has been declined by your `auto-load safe-path' set to
"$debugdir:$datadir/auto-load".
To enable execution of this file add
add-auto-load-safe-path
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/gcc/.gdbinit
line to your configuration file "/home/snoonan/.gdbinit".
To completely disable this security protection add
set auto-load safe-path /
line to your configuration file "/home/snoonan/.gdbinit".
For more information about this security protection see the
"Auto-loading safe path" section in the GDB manual.  E.g., run from the shell:
info "(gdb)Auto-loading safe path"
(gdb) set follow-fork-mode child
(gdb) run
Starting program:
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/prev-gcc/xgcc
-B/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/./prev-gcc/
-B/usr/x86_64-unknown-linux-gnu/bin/ -B/usr/x86_64-unknown-linux-gnu/bin/
-B/usr/x86_64-unknown-linux-gnu/lib/ -isystem
/usr/x86_64-unknown-linux-gnu/include -isystem
/usr/x86_64-unknown-linux-gnu/sys-include -c -g -O2 -flto=jobserver
-frandom-seed=1 -gnatpg -W -Wall -nostdinc -I- -I. -Iada/generated -Iada
-I/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/ada
-I/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/ada/gcc-interface
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/ada/comperr.adb
-o ada/comperr.o
[New process 16017]
process 16017 is executing new program:
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-build/prev-gcc/gnat1

Program received signal SIGSEGV, Segmentation fault.
[Switching to process 16017]
0x00c354ed in ei_edge (i=...) at
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/basic-block.h:460
460   return EDGE_I (ei_container (i), i.index);
(gdb) thread apply all bt full

Thread 2 (process 16017):
#0  0x00c354ed in ei_edge (i=...) at
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/basic-block.h:460
No locals.
#1  0x00c376cf in inverted_post_order_compute (post_order=0x2e3d9a0,
start_points=0x2bb0858 ) at
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/cfganal.c:837
ei = {
  index = 0, 
  container = 0x768543a8
}
pred = 0x76c8b8be 
has_unvisited_bb = false
bb = 0x0
stack = 0x2e1e740
sp = 171
post_order_num = 0
visited = 0x2decb40
__FUNCTION__ = "inverted_post_order_compute"
#2  0x011ddcdc in remove_dead_stmt (i=0x7fffc950,
bb=0x75ad07b8) at
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/tree-ssa-dce.c:1076
postorder = 0x2e3d9a0
postorder_num = 0
ei = {
  index = 4121675328, 
  container = 0x7fffc860
}
e = 0x0
e2 = 0xdc9474 
stmt = 0x75abc640
__FUNCTION__ = "remove_dead_stmt"
#3  0x011de918 in eliminate_unnecessary_stmts () at
/home/snoonan/Development/ec2-packages/gcc-multilib/src/gcc-5-20150602/gcc/tree-ssa-dce.c:1360
something_changed = true
  

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[hubicka at ucw dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

--- Comment #8 from Jan Hubicka  ---
 [...]
> raised STORAGE_ERROR : stack overflow or erroneous memory access  
> /build/gcc-multilib/src/gcc-5-20150602/gcc/ada/gcc-interface/Make-lang.in:119:
> recipe for target 'ada/comperr.o' failed
> 
> Ideas?

This is most likely an independent issue.  Can you, please, attach a backtrace?

Honza

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[steven at uplinklabs dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

--- Comment #7 from Steven Noonan  ---
Tried applying the patch mentioned in comment 6 and doing a build using
--with-build-config=bootstrap-lto. Ended with:

[...]
/build/gcc-multilib/src/gcc-build/./prev-gcc/xgcc
-B/build/gcc-multilib/src/gcc-build/./prev-gcc/
-B/usr/x86_64-unknown-linux-gnu/bin
/ -B/usr/x86_64-unknown-linux-gnu/bin/ -B/usr/x86_64-unknown-linux-gnu/lib/
-isystem /usr/x86_64-unknown-linux-gnu/include -isystem /
usr/x86_64-unknown-linux-gnu/sys-include-c -g -O2 -flto=jobserver
-frandom-seed=1  -gnatpg  -W -Wall -nostdinc -I- -I. -Iada/gene
rated -Iada -I/build/gcc-multilib/src/gcc-5-20150602/gcc/ada
-I/build/gcc-multilib/src/gcc-5-20150602/gcc/ada/gcc-interface /build/gc
c-multilib/src/gcc-5-20150602/gcc/ada/comperr.adb -o ada/comperr.o
[...]
raised STORAGE_ERROR : stack overflow or erroneous memory access  
/build/gcc-multilib/src/gcc-5-20150602/gcc/ada/gcc-interface/Make-lang.in:119:
recipe for target 'ada/comperr.o' failed

Ideas?

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

--- Comment #6 from Jan Hubicka  ---
Patch posted to https://gcc.gnu.org/ml/gcc-patches/2015-05/msg02876.html

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

Jan Hubicka  changed:

   What|Removed |Added

   Last reconfirmed|2015-03-13 00:00:00 |2015-4-30

--- Comment #5 from Jan Hubicka  ---
Still ICE during LTObootstrap on PPC64 (gcc110). I will try to at least figure
out what goes wrong with the PHI (if it is wrong control dependency due to
critical edge or something else)

I suppose we can relax the check to:

  degenerate_phi_p (phi) || SSA_NAME_OCCURS_IN_ABNORMAL_PHI (gimple_phi_result
(phi)

but we may need to trigger SSA rewrite on resulting code as cddce will break
dominance conditions on the PHI.

[Bug tree-optimization/65337] [5/6 Regression] bootstrap-lto gnat1 linktime ICE: gcc/ada/exp_aggr.adb:6570:0: internal compiler error: in forward_edge_to_pdom, at tree-ssa-dce.c:1086

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65337

Jakub Jelinek  changed:

   What|Removed |Added

   Target Milestone|5.0 |5.2

--- Comment #4 from Jakub Jelinek  ---
GCC 5.1 has been released.