[Bug debug/42977] New: "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[Bug debug/42977] New: "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[zsojka at seznam dot cz]

Command line:
g++ -O2 -finline-functions -fomit-frame-pointer -ftracer
-fsched2-use-superblocks -fPIC -fcompare-debug -c testcase.cpp -m32
(x86_64 needs -m32)

Tested revisions:
r156515 - crash
r156472 - crash
r156468 - crash
r156467 - OK

Testcase before reducing was crashing stopped crashing in r156468. Some
partially-reduced testcases were crashing in different revisions. It can happen
this will stop crashing in future revisions because it will hide again.


-- 
           Summary: "-fcompare-debug failure" with "-O2 -finline-functions -
                    fomit-frame-pointer -ftracer -fsched2-use-superblocks -
                    fPIC"
           Product: gcc
           Version: 4.5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: debug
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: zsojka at seznam dot cz
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[zsojka at seznam dot cz]

------- Comment #1 from zsojka at seznam dot cz  2010-02-05 18:14 -------
Created an attachment (id=19809)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=19809&action=view)
most reduced testcase, still quite long

Command line:
g++ -O2 -finline-functions -fomit-frame-pointer -ftracer
-fsched2-use-superblocks -fPIC -fcompare-debug -m32 -c testcase2.cpp


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|"-fcompare-debug failure"   |[4.5 Regression] "-fcompare-
                   |with "-O2 -finline-functions|debug failure" with "-O2 -
                   |-fomit-frame-pointer -      |finline-functions -fomit-
                   |ftracer -fsched2-use-       |frame-pointer -ftracer -
                   |superblocks -fPIC"          |fsched2-use-superblocks -
                   |                            |fPIC"
   Target Milestone|---                         |4.5.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[rguenth at gcc dot gnu dot org]

------- Comment #2 from rguenth at gcc dot gnu dot org  2010-02-07 15:06 -------
Confirmed.  The difference is from scheduling:

--- t.gk.gkd    2010-02-07 16:05:02.000000000 +0100
+++ t.gkd       2010-02-07 16:05:01.000000000 +0100
@@ -1900,6 +1900,9 @@
 (insn:TI# 0 0 t.C:403 (set (reg/f:SI 0 ax [389])
         (plus:SI (reg/f:SI 7 sp)
             (const_int 79 [0x4f])))# {*lea_1} (nil))
+(insn:TI# 0 0 t.C:403 (set (mem:SI (reg/f:SI 7 sp) [0 S4 A32])
+        (reg/f:SI 0 ax [389]))# {*movsi_1} (expr_list:REG_DEAD (reg/f:SI 0 ax
[
389])
+        (nil)))
 (insn:TI# 0 0 t.C:403 (set (mem/f/i:SI (plus:SI (reg/f:SI 7 sp)
                 (const_int 8 [0x8])) [0 S4 A32])
         (reg/f:SI 4 si [517]))# {*movsi_1} (expr_list:REG_EQUAL (plus:SI
(reg/f
:SI 20 frame)
@@ -1911,9 +1914,6 @@
         (expr_list:REG_EQUAL (plus:SI (reg/f:SI 20 frame)
                 (const_int -5 [0xfffffffffffffffb]))
             (nil))))
-(insn:TI# 0 0 t.C:403 (set (mem:SI (reg/f:SI 7 sp) [0 S4 A32])
-        (reg/f:SI 0 ax [389]))# {*movsi_1} (expr_list:REG_DEAD (reg/f:SI 0 ax
[
389])
-        (nil)))
 (call_insn:TI# 0 0 t.C:403 (parallel [
             (call (mem:QI (symbol_ref:SI
("_ZN4Glib12SignalProxy2Iv7ustringS1_E
7connectEN4sigc4slotIvNS3_3nilES5_S5_S5_EE") [flags 0x41]  <function_decl #
conn
ect>) [0 S1 A8])
                 (const_int 12 [0xc]))


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
           Priority|P3                          |P1
   Last reconfirmed|0000-00-00 00:00:00         |2010-02-07 15:06:03
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[jakub at gcc dot gnu dot org]

------- Comment #3 from jakub at gcc dot gnu dot org  2010-02-09 13:41 -------
This is because the [sp] = ax insn has different INSN_PRIORITY between -g and
-g0.  That difference is because different kind of dependency is added when
analyzing the following call:
(call (mem:QI (symbol_ref:SI
("_ZN4Glib12SignalProxy2Iv7ustringS1_E7connectEN4sigc4slotIvNS3_3nilES5_S5_S5_EE")
[flags 0x41] <function_decl 0x7ffff17bf700 connect>) [0 S1 A8])
    (const_int 12 [0xc]))
Without -g the (mem (value)) in question has 2 different locs:
p *mem->u.fld[0].rt_rtx->u.fld[0].rt_cselib->locs
{next = 0x1a57510, loc = 0x7ffff1822460, setting_insn = 0x7ffff13c1360}
debug_rtx (mem->u.fld[0].rt_rtx->u.fld[0].rt_cselib->locs->loc)
(reg/f:SI 7 sp)
debug_rtx (mem->u.fld[0].rt_rtx->u.fld[0].rt_cselib->locs->next->loc)
(plus:SI (value:SI 274:4212 @0x1a4ffb8/0x1a57790)
    (const_int -4 [0xfffffffffffffffc]))
while with -g there is just one loc - sp.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[jakub at gcc dot gnu dot org]

------- Comment #4 from jakub at gcc dot gnu dot org  2010-02-09 14:34 -------
Argh, this isn't going to be fun to fix.
The problem is:
2018  if (n_useless_values > MAX_USELESS_VALUES
2019      /* remove_useless_values is linear in the hash table size.  Avoid
2020         quadratic behavior for very large hashtables with very few
2021         useless elements.  */
2022      && (unsigned int)n_useless_values > cselib_hash_table->n_elements /
4)
2023    remove_useless_values ();

in cselib_process_insn.  As scheduling (but I bet other cselib users too,
though no idea if it can make a difference in that case) does call
cselib_process_insn
even on DEBUG_INSNs, n_useless_values and/or cselib_hash_table->n_elements can
differ between -g and -g0 compilations.  Either we'd need to ensure this
compaction is done at the same times (track number of useless values caused by
processing non-DEBUG_INSNs and similarly elements), or get_addr would need to
ignore useless, but not yet removed as useless, VALUEs, or something similar.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[matz at gcc dot gnu dot org]

------- Comment #5 from matz at gcc dot gnu dot org  2010-03-19 16:59 -------
How about not calling cselib_process_insn on DEBUG_INSNs (or better make
cselib_process_insn ignore them).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[jakub at gcc dot gnu dot org]

------- Comment #6 from jakub at gcc dot gnu dot org  2010-03-19 17:01 -------
Well, we have to call cselib_process_insn on DEBUG_INSNs at least during
var-tracking.  As for other passes, such as DSE, haven't thought about the
impliciations of not doing so.  Alex?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[jakub at gcc dot gnu dot org]

------- Comment #7 from jakub at gcc dot gnu dot org  2010-03-22 10:12 -------
I believe e.g. DSE, postreload (reload_cse_regs_1), cfgcleanup (thread_jump)
wouldn't mind guarding cselib_process_insn call with
if (!DEBUG_INSN_P (insn))
  cselib_process_insn (insn);
On the other side, e.g. ebb scheduling would care a lot, it uses cselib for
MEMs
both in normal insns and in debug insns to find out if a move is possible and
in case of debug insns if the DEBUG insn doesn't need to be dropped.
I believe gcse (local_cprop_pass) would also care.  So we have 3 passes that
wouldn't care, 2 that would and one (var-tracking) that must handle
DEBUG_INSNs, but where it never can't create any -fcompare-debug failures,
simply because
var-tracking only affects var_location notes and not code and is only run in
the -g case.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[jakub at gcc dot gnu dot org]

------- Comment #8 from jakub at gcc dot gnu dot org  2010-03-22 15:21 -------
BTW, when not using cselib_record_sets hook, I believe cselib_process_insn does
really nothing at all on DEBUG_INSNs.  This would also mean that even local
cprop is fine.  The problem are cselib_lookup (x, y, 1); calls sched-deps.c
does
on MEMs in debug_insns.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[matz at gcc dot gnu dot org]

------- Comment #9 from matz at gcc dot gnu dot org  2010-03-22 16:01 -------
For sched-deps.c there are two calls to cselib_lookup:
1) in sched_analize_1 (for writing to MEM)
2) in sched_analize_2 (for reading a MEM)

Regarding (2): cselib_lookup is called on a copy of X (made into T) which
then is replaced into XEXP(t,0).  But for debug-insn that very T isn't used
at all further down.  So we can disregard the use at (2) (and move the whole
manipulation of T into the !DEBUG_INSN_P block).

Regarding (1): Here we don't handle debug-insns specially, but sched_analize_1
is called only on PRE/POST_DEC/INC/MODIFY, top-level SET/CLOBBER insns (maybe
in a PARALLEL) or CLOBBERS in CALL_INSN_FUNCTION_USAGE.

It better should be true that debug-insn don't contain a
PRE/POST_DEC/INC/MODIFY side-effect, or a top-level SET/CLOBBER.  IOW
sched_analyze_1 never should be called on debug-insns.  So we can disregard
case (1) too.

Next: if the cselib_lookup calls (that always have their 'create' argument
be true) ever create a new VALUE for a MEM in a debug-insn, that same wouldn't
happen without debug-insn, or in different order.  I think this would lead
to debug-miscompare later down anyway.

So, I think any calls into cselib on debug-insns that happen to create new
values are actually bugs waiting to be discovered.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[jakub at gcc dot gnu dot org]

------- Comment #11 from jakub at gcc dot gnu dot org  2010-03-30 21:01 -------
Fixed.


-- 

jakub at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977

[Bug debug/42977] [4.5 Regression] "-fcompare-debug failure" with "-O2 -finline-functions -fomit-frame-pointer -ftracer -fsched2-use-superblocks -fPIC"

[jakub at gcc dot gnu dot org]

------- Comment #10 from jakub at gcc dot gnu dot org  2010-03-30 21:01 -------
Subject: Bug 42977

Author: jakub
Date: Tue Mar 30 21:00:47 2010
New Revision: 157837

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=157837
Log:
        PR debug/42977
        * cselib.c (n_useless_values): Document handling of debug locs.
        (n_useless_debug_values, n_debug_values): New variables.
        (new_elt_loc_list): Don't add to debug values, keep count.
        (promote_debug_loc): New.
        (cselib_reset_table): Zero new variables.
        (entry_and_rtx_equal_p): Promote debug locs.
        (discard_useless_locs): Increment n_useless_debug_values for
        debug values.
        (remove_useless_values): Adjust n_useless_values and n_debug_values
        with n_useless_debug_values.
        (add_mem_for_addr): Promote debug locs.
        (cselib_lookup_mem): Likewise.
        (cselib_lookup_addr): Renamed to...
        (cselib_lookup_addr_1): ... this.  Promote debug locs.
        Don't call...
        (cselib_log_lookup): ... this.  Turn into...
        (cselib_lookup_addr): ... new wrapper.
        (cselib_lookup_from_insn): New.
        (cselib_invalidate_regno): Increment n_useless_debug_values for
        debug values.
        (cselib_invalidate_mem): Likewise.
        (cselib_process_insn): Take n_deleted and n_debug_values into
        account to guard remove_useless_value call.
        (cselib_finish): Zero n_useless_debug_values.
        * cselib.h (cselib_lookup_from_insn): Declare.
        * sched-deps.c (sched_analyze_1): Use cselib_lookup_from_insn.
        (sched_analyze_2): Likewise.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/cselib.c
    trunk/gcc/cselib.h
    trunk/gcc/sched-deps.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42977