[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 Richard Biener changed: What|Removed |Added Status|REOPENED|RESOLVED Resolution|--- |FIXED --- Comment #21 from Richard Biener --- Thus fixed.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 Arseny Solokha changed: What|Removed |Added CC||asolokha at gmx dot com --- Comment #20 from Arseny Solokha --- As it was fixed on trunk in 2012, it's actually fixed for every supported release branch in 2016. Indeed, it's not reproducible from gcc 4.8 on anymore.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 Aldy Hernandez aldyh at gcc dot gnu.org changed: What|Removed |Added Status|NEW |RESOLVED Resolution||FIXED --- Comment #18 from Aldy Hernandez aldyh at gcc dot gnu.org 2012-11-28 20:56:44 UTC --- fixed on trunk
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 Aldy Hernandez aldyh at gcc dot gnu.org changed: What|Removed |Added Status|RESOLVED|REOPENED Resolution|FIXED | --- Comment #19 from Aldy Hernandez aldyh at gcc dot gnu.org 2012-11-28 20:58:06 UTC --- My bad. Didn't notice this was for 4.7. Reopening.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 --- Comment #16 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-24 08:57:12 UTC --- Author: rguenth Date: Mon Sep 24 08:57:08 2012 New Revision: 191658 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=191658 Log: 2012-09-24 Richard Guenther rguent...@suse.de PR middle-end/52173 * gimple.c (gimple_copy): Properly mark the copy modified if SSA operands are present. * gcc.dg/tm/pr52173-1.c: New. * gcc.dg/tm/pr52173-2.c: New. Added: trunk/gcc/testsuite/gcc.dg/tm/pr52173-1.c trunk/gcc/testsuite/gcc.dg/tm/pr52173-2.c Modified: trunk/gcc/ChangeLog trunk/gcc/gimple.c trunk/gcc/testsuite/ChangeLog
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 Richard Guenther rguenth at gcc dot gnu.org changed: What|Removed |Added Status|ASSIGNED|NEW Known to work||4.8.0 AssignedTo|rguenth at gcc dot gnu.org |unassigned at gcc dot ||gnu.org Known to fail||4.7.2 --- Comment #17 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-24 08:58:20 UTC --- Fixed on trunk.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 Richard Guenther rguenth at gcc dot gnu.org changed: What|Removed |Added CC||srdjan.stipic at gmail dot ||com --- Comment #12 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-21 08:55:25 UTC --- *** Bug 54646 has been marked as a duplicate of this bug. ***
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173
--- Comment #13 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-21
12:39:19 UTC ---
Simpler testcase from PR54646, fails at -O2:
int a;
int main()
{
int i;
for (i = 0; i 1; ++i)
__transaction_atomic { ++a; }
return 0;
}
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173
--- Comment #14 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-21
13:27:30 UTC ---
We have broken immediate uses of .MEM_2 (for example). Before update-ssa:
main ()
{
int a.1;
int a.0;
int i;
bb 2:
bb 8:
# i_10 = PHI 0(2)
# .MEM_11 = PHI .MEM_3(D)(2)
bb 9:
# .MEM_4 = VDEF .MEM_2
__transaction_atomic // SUBCODE=[ GTMA_HAVE_LOAD GTMA_HAVE_STORE ]
...
bb 3:
# .MEM_12 = VDEF .MEM_2
__transaction_atomic // SUBCODE=[ GTMA_HAVE_LOAD GTMA_HAVE_STORE ]
...
bb 7:
# VUSE .MEM_2
return 0;
but debug_immediate_uses shows:
.MEM_2 : --2 uses.
# VUSE .MEM_2
return 0;
# .MEM_12 = VDEF .MEM_2
__transaction_atomic // SUBCODE=[ GTMA_HAVE_LOAD GTMA_HAVE_STORE ]
that means that update_stmt didn't do its job or it wasn't called.
Aa. Because in gimple_copy:
/* Make copy of operands. */
if (num_ops 0)
{
...
/* SSA operands need to be updated. */
gimple_set_modified (copy, true);
}
there are zero (real) operands but we _do_ have virtual ops.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173
Richard Guenther rguenth at gcc dot gnu.org changed:
What|Removed |Added
Status|NEW |ASSIGNED
AssignedTo|aldyh at gcc dot gnu.org|rguenth at gcc dot gnu.org
--- Comment #15 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-21
13:30:06 UTC ---
I have a patch:
Index: gcc/gimple.c
===
*** gcc/gimple.c(revision 191613)
--- gcc/gimple.c(working copy)
*** gimple_copy (gimple stmt)
*** 2333,2355
}
/* Make copy of operands. */
! if (num_ops 0)
! {
! for (i = 0; i num_ops; i++)
! gimple_set_op (copy, i, unshare_expr (gimple_op (stmt, i)));
! /* Clear out SSA operand vectors on COPY. */
! if (gimple_has_ops (stmt))
! {
! gimple_set_def_ops (copy, NULL);
! gimple_set_use_ops (copy, NULL);
! }
! if (gimple_has_mem_ops (stmt))
! {
! gimple_set_vdef (copy, gimple_vdef (stmt));
! gimple_set_vuse (copy, gimple_vuse (stmt));
! }
/* SSA operands need to be updated. */
gimple_set_modified (copy, true);
--- 2333,2352
}
/* Make copy of operands. */
! for (i = 0; i num_ops; i++)
! gimple_set_op (copy, i, unshare_expr (gimple_op (stmt, i)));
! if (gimple_has_mem_ops (stmt))
! {
! gimple_set_vdef (copy, gimple_vdef (stmt));
! gimple_set_vuse (copy, gimple_vuse (stmt));
! }
! /* Clear out SSA operand vectors on COPY. */
! if (gimple_has_ops (stmt))
! {
! gimple_set_def_ops (copy, NULL);
! gimple_set_use_ops (copy, NULL);
/* SSA operands need to be updated. */
gimple_set_modified (copy, true);
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173
--- Comment #10 from rguenther at suse dot de rguenther at suse dot de
2012-09-20 07:43:56 UTC ---
On Wed, 19 Sep 2012, aldyh at gcc dot gnu.org wrote:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173
Aldy Hernandez aldyh at gcc dot gnu.org changed:
What|Removed |Added
CC||amacleod at redhat dot com,
||dnovillo at gcc dot
||gnu.org, rguenth at gcc dot
||gnu.org
--- Comment #9 from Aldy Hernandez aldyh at gcc dot gnu.org 2012-09-19
19:44:14 UTC ---
Richi, Diego. Perhaps you can shed light on this.
Here we have a loop unroll that triggers a virtual def operand missing...
for
a GIMPLE_TRANSACTION:
int vec[500];
void func()
{
__transaction_relaxed
{
vec[123] = 456;
}
}
main()
{
int i;
for(i = 0; i 10; ++i)
func();
}
The unroller wants to unroll into something like:
bb 2:
# .MEM_6 = VDEF .MEM_13
__transaction_relaxed-- PROBLEMATIC INSN (VDEF's)
bb 3:
# .MEM_9 = VDEF .MEM_59
vec[123] = 456;
# .MEM_2 = VDEF .MEM_9
__builtin__ITM_commitTransaction ();
bb 4:
ivtmp_14 = 9;
# .MEM_11 = VDEF .MEM_13
__transaction_relaxed
bb 5:
# .MEM_18 = VDEF .MEM_59
vec[123] = 456;
# .MEM_19 = VDEF .MEM_18
__builtin__ITM_commitTransaction ();
etc etc etc.
Putting aside how incredibly inefficient this is... (We should be duplicating
the vector store, not the entire transaction)...
The problem here is that, after unrolling, the VDEF of the problematic insn is
.MEM_6, but the DEF_OP of the insn is .MEM_59. So gimple_vdef() !=
gimple_vdef_op() and we trigger a virtual def operand missing for stmt.
The problem is that the unroller's call to gimple_duplicate_bb() will make a
copy of the problematic instruction, maintaining its gimple_vdef, but changing
all its def_ops from under us, thus making gimple_vdef() != gimple_vdef_op().
Setting a DEF will change the vdef. If it does not, the stmt was
not updated properly.
Before loop unrolling we have this snippet that is about to be unrolled:
bb 3:
# .MEM_13 = PHI .MEM_8(6), .MEM_3(D)(2)
# ivtmp_1 = PHI ivtmp_10(6), 10(2)
# .MEM_6 = VDEF .MEM_13
__transaction_relaxed
After loop unrolling the above remains unchanged, but the following is
inserted
*before* bb 3. [Note, the following is after update_ssa(TODO_update_ssa), but
before cleanup_tree_cfg() for clarity.]
bb 8:
# .MEM_12 = PHI .MEM_3(D)(2)
# ivtmp_5 = PHI 10(2)
# .MEM_6 = VDEF .MEM_13 -- *** --
-- shouldn't this be .MEM_12 ?
-- *** --
Yes.
__transaction_relaxed
...
...more unrolling
...
bb 40:
# .MEM_57 = PHI .MEM_8(39)
# ivtmp_58 = PHI ivtmp_10(39)
# .MEM_53 = VDEF .MEM_13-- ** --
-- shouldn't this be .MEM_57
-- ** --
Yes.
__transaction_relaxed
...
...
bb 3
..
Notice that both BB8 and BB40 invalidate MEM_13 in its virtual definition.
Shouldn't the VDEF.MEM_13 be rewritten as I have indicated above?
Especially
since now .MEM_13 is defined much further down in bb 3:
bb 8
# invalidates VDEF.MEM_13
...
bb 40
# invalidates VDEF.MEM_13
..
bb 3
# define .MEM_13 =
It seems gimple_duplicate_bb() renames all the phis and the operands, but does
not update the VDEF's for the transaction. Gimple_duplicate_bb() says it
doesn't maintain SSA form, so I assume this in on purpose (?). So how is this
supposed to be cleaned up? Or is this even the problem?
The problem must be in the IL before unrolling I believe.
Richard.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 --- Comment #11 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-20 15:08:55 UTC --- (In reply to comment #10) On Wed, 19 Sep 2012, aldyh at gcc dot gnu.org wrote: After loop unrolling the above remains unchanged, but the following is inserted *before* bb 3. [Note, the following is after update_ssa(TODO_update_ssa), but before cleanup_tree_cfg() for clarity.] bb 8: # .MEM_12 = PHI .MEM_3(D)(2) # ivtmp_5 = PHI 10(2) # .MEM_6 = VDEF .MEM_13 -- *** -- -- shouldn't this be .MEM_12 ? -- *** -- Yes. Btw, is_new_name says yes to .MEM_12, false for .MEM_6 and 13. update_ssa should have adjusted the use .MEM_13 to .MEM_12, but it didn't for some reason. I see that the DEF stmt for .MEM_6 is # .MEM_59 = VDEF .MEM_13 __transaction_relaxed // SUBCODE=[ GTMA_HAVE_STORE ] though before update_ssa. That looks bogus to me. Now I have to leave ... maybe the above is some hints to you (or I will try to look further tomorrow).
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173
Aldy Hernandez aldyh at gcc dot gnu.org changed:
What|Removed |Added
CC||amacleod at redhat dot com,
||dnovillo at gcc dot
||gnu.org, rguenth at gcc dot
||gnu.org
--- Comment #9 from Aldy Hernandez aldyh at gcc dot gnu.org 2012-09-19
19:44:14 UTC ---
Richi, Diego. Perhaps you can shed light on this.
Here we have a loop unroll that triggers a virtual def operand missing... for
a GIMPLE_TRANSACTION:
int vec[500];
void func()
{
__transaction_relaxed
{
vec[123] = 456;
}
}
main()
{
int i;
for(i = 0; i 10; ++i)
func();
}
The unroller wants to unroll into something like:
bb 2:
# .MEM_6 = VDEF .MEM_13
__transaction_relaxed-- PROBLEMATIC INSN (VDEF's)
bb 3:
# .MEM_9 = VDEF .MEM_59
vec[123] = 456;
# .MEM_2 = VDEF .MEM_9
__builtin__ITM_commitTransaction ();
bb 4:
ivtmp_14 = 9;
# .MEM_11 = VDEF .MEM_13
__transaction_relaxed
bb 5:
# .MEM_18 = VDEF .MEM_59
vec[123] = 456;
# .MEM_19 = VDEF .MEM_18
__builtin__ITM_commitTransaction ();
etc etc etc.
Putting aside how incredibly inefficient this is... (We should be duplicating
the vector store, not the entire transaction)...
The problem here is that, after unrolling, the VDEF of the problematic insn is
.MEM_6, but the DEF_OP of the insn is .MEM_59. So gimple_vdef() !=
gimple_vdef_op() and we trigger a virtual def operand missing for stmt.
The problem is that the unroller's call to gimple_duplicate_bb() will make a
copy of the problematic instruction, maintaining its gimple_vdef, but changing
all its def_ops from under us, thus making gimple_vdef() != gimple_vdef_op().
Before loop unrolling we have this snippet that is about to be unrolled:
bb 3:
# .MEM_13 = PHI .MEM_8(6), .MEM_3(D)(2)
# ivtmp_1 = PHI ivtmp_10(6), 10(2)
# .MEM_6 = VDEF .MEM_13
__transaction_relaxed
After loop unrolling the above remains unchanged, but the following is inserted
*before* bb 3. [Note, the following is after update_ssa(TODO_update_ssa), but
before cleanup_tree_cfg() for clarity.]
bb 8:
# .MEM_12 = PHI .MEM_3(D)(2)
# ivtmp_5 = PHI 10(2)
# .MEM_6 = VDEF .MEM_13 -- *** --
-- shouldn't this be .MEM_12 ?
-- *** --
__transaction_relaxed
...
...more unrolling
...
bb 40:
# .MEM_57 = PHI .MEM_8(39)
# ivtmp_58 = PHI ivtmp_10(39)
# .MEM_53 = VDEF .MEM_13-- ** --
-- shouldn't this be .MEM_57
-- ** --
__transaction_relaxed
...
...
bb 3
..
Notice that both BB8 and BB40 invalidate MEM_13 in its virtual definition.
Shouldn't the VDEF.MEM_13 be rewritten as I have indicated above? Especially
since now .MEM_13 is defined much further down in bb 3:
bb 8
# invalidates VDEF.MEM_13
...
bb 40
# invalidates VDEF.MEM_13
..
bb 3
# define .MEM_13 =
It seems gimple_duplicate_bb() renames all the phis and the operands, but does
not update the VDEF's for the transaction. Gimple_duplicate_bb() says it
doesn't maintain SSA form, so I assume this in on purpose (?). So how is this
supposed to be cleaned up? Or is this even the problem?
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 --- Comment #8 from Aldy Hernandez aldyh at gcc dot gnu.org 2012-09-14 21:20:23 UTC --- Created attachment 28195 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=28195 reduced testcase Further reduced testcase that triggers on either C/C++.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 --- Comment #6 from Aldy Hernandez aldyh at gcc dot gnu.org 2012-09-12 19:52:55 UTC --- I'll take a look.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 --- Comment #6 from Aldy Hernandez aldyh at gcc dot gnu.org 2012-09-12 19:52:55 UTC --- I'll take a look. --- Comment #7 from Aldy Hernandez aldyh at gcc dot gnu.org 2012-09-12 19:53:30 UTC --- I'll take a look.
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 Jakub Jelinek jakub at gcc dot gnu.org changed: What|Removed |Added Status|UNCONFIRMED |NEW Last reconfirmed||2012-08-13 CC||aldyh at gcc dot gnu.org, ||jakub at gcc dot gnu.org Ever Confirmed|0 |1
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173
John Regehr regehr at cs dot utah.edu changed:
What|Removed |Added
CC||regehr at cs dot utah.edu
--- Comment #3 from John Regehr regehr at cs dot utah.edu 2012-02-15 03:50:18
UTC ---
Smaller testcase in pure C for apparently the same problem.
[regehr@gamow 1]$ current-gcc -Ofast small.c
small.c: In function 'f':
small.c:3:6: error: definition in block 8 does not dominate use in block 9
for SSA_NAME: .MEM_22 in statement:
# .MEM_29 = VDEF .MEM_22
c[0] = c_I_lsm.9_32;
small.c:3:6: internal compiler error: verify_ssa failed
Please submit a full bug report,
with preprocessed source if appropriate.
See http://gcc.gnu.org/bugs.html for instructions.
[regehr@gamow 1]$ cat small.c
int a, b, c[10], d[10] = { 0, 0 };
void f (void) {
for (a = 1; a = 4; a += 1)
d[a] = d[1];
for (; b; ++b)
c[0] |= 1;
}
[regehr@gamow 1]$ current-gcc -v
Using built-in specs.
COLLECT_GCC=current-gcc
COLLECT_LTO_WRAPPER=/uusoc/exports/scratch/regehr/z/compiler-install/gcc-r184237-install/bin/../libexec/gcc/x86_64-unknown-linux-gnu/4.7.0/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: ../configure --with-libelf=/usr/local --enable-lto
--prefix=/home/regehr/z/compiler-install/gcc-r184237-install
--program-prefix=r184237- --enable-languages=c,c++
Thread model: posix
gcc version 4.7.0 20120214 (experimental) (GCC)
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 --- Comment #4 from Andrew Pinski pinskia at gcc dot gnu.org 2012-02-15 03:53:52 UTC --- (In reply to comment #3) Smaller testcase in pure C for apparently the same problem. This is a different issue, Can you file a bug report about this one?
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 --- Comment #5 from John Regehr regehr at cs dot utah.edu 2012-02-15 03:58:22 UTC --- Sorry, see PR52255
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 --- Comment #1 from Jack Lloyd lloyd at randombit dot net 2012-02-08 15:52:19 UTC --- Created attachment 26615 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26615 Testcase
[Bug middle-end/52173] internal compiler error: verify_ssa failed possibly caused by itm
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52173 Jack Lloyd lloyd at randombit dot net changed: What|Removed |Added Attachment #26615|0 |1 is obsolete|| --- Comment #2 from Jack Lloyd lloyd at randombit dot net 2012-02-08 15:54:25 UTC --- Created attachment 26616 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26616 Testcase Elide the thread stuff from the testcase, doesn't affect the error
