[Bug tree-optimization/44258] New: possible SRA wrong-code generation.

[Bug tree-optimization/44258] New: possible SRA wrong-code generation.

[pluto at agmk dot net]

the recent gcc-4.5-branch produces broken libkhtml.so.5.4.0 from kdelibs-4.4.3.
afaics it produces different/broken binaries for khtml/css/parser.cpp
and khtml/svg/SVGGradientElement.cpp.

finally we get a nice GPF during knode/kmail/konqueror startup:

[KCrash Handler]
#5  memcpy () at ../sysdeps/x86_64/memcpy.S:78
#6  0x00007f546e63fc5e in QString::QString(QChar const*, int) () from 
/usr/lib64/libQtCore.so.4
#7  0x00007f5469f70e2e in qString (ps=<value optimized out>) at 
/usr/src/debug/kdelibs-4.4.3/khtml/css/cssparser.h:84
#8  DOM::CSSParser::parseValue (ps=<value optimized out>) at 
/usr/src/debug/kdelibs-4.4.3/khtml/css/cssparser.cpp:518
#9  0x00007f5469f95075 in cssyyparse (parser=0x7fff08c22820) at 
/usr/src/debug/kdelibs-4.4.3/khtml/css/parser.cpp:2969
#10 0x00007f5469f67d00 in DOM::CSSParser::runParser (this=0x7fff08c22820) at 
/usr/src/debug/kdelibs-4.4.3/khtml/css/cssparser.cpp:151
(...)

with svn-bisect i've identified the first affected revision:

Author: jamborm
Date: Wed Apr 28 13:09:56 2010
New Revision: 158826

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=158826
Log:
2010-04-28  Martin Jambor  <mjambor@suse.cz>

        PR tree-optimization/43846
        * tree-sra.c (struct access): New flag grp_assignment_read.
        (build_accesses_from_assign): Set grp_assignment_read.
        (sort_and_splice_var_accesses): Propagate grp_assignment_read.
        (enum mark_read_status): New type.
        (analyze_access_subtree): Propagate grp_assignment_read, create
        accesses also if both direct_read and root->grp_assignment_read.

        * testsuite/gcc.dg/tree-ssa/sra-10.c: New test.


-- 
           Summary: possible SRA wrong-code generation.
           Product: gcc
           Version: 4.5.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pluto at agmk dot net
 GCC build triplet: x86_64-gnu-linux
  GCC host triplet: x86_64-gnu-linux
GCC target triplet: x86_64-gnu-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] possible SRA wrong-code generation.

[pluto at agmk dot net]

------- Comment #1 from pluto at agmk dot net  2010-05-24 11:02 -------
Created an attachment (id=20732)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20732&action=view)
preprocessed parser from kdelibs sources.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |4.5.1
      Known to work|                            |4.5.0
            Summary|possible SRA wrong-code     |[4.5/4.6 Regression]
                   |generation.                 |possible SRA wrong-code
                   |                            |generation.
   Target Milestone|---                         |4.5.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[arekm at pld-linux dot org]

------- Comment #2 from arekm at pld-linux dot org  2010-05-24 12:14 -------
In meantime - is reversing the problematic gcc commit a sane thing to do for a
gcc user? (from what I understand it was simply a better optimization and no
real bugfix, right?)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[jamborm at gcc dot gnu dot org]

------- Comment #3 from jamborm at gcc dot gnu dot org  2010-05-24 12:57 -------
(In reply to comment #2)
> In meantime - is reversing the problematic gcc commit a sane thing to do for a
> gcc user? (from what I understand it was simply a better optimization and no
> real bugfix, right?)
> 

If you're asking about reverting it in your local copy and by sane you
mean safe then yes, you certainly can do it.  I'll have a look at this
as soon as I can but unfortunately that probably means next week.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
           Keywords|                            |wrong-code
           Priority|P3                          |P1
   Last reconfirmed|0000-00-00 00:00:00         |2010-05-30 17:04:29
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[jamborm at gcc dot gnu dot org]

------- Comment #4 from jamborm at gcc dot gnu dot org  2010-06-03 09:30 -------
Unfortunately the preprocessed source from comment #1 seems to be
damaged, I get loads of errors like "error: stray '\336' in program."
Can you please re-upload it?  Thanks.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[torbenh at users dot sourceforge dot net]

------- Comment #5 from torbenh at users dot sourceforge dot net  2010-06-03 19:15 -------
(In reply to comment #4)
> Unfortunately the preprocessed source from comment #1 seems to be
> damaged, I get loads of errors like "error: stray '\336' in program."
> Can you please re-upload it?  Thanks.
> 

have a look at PR44406... i believe its the same thing.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[maksim at kde dot org]

------- Comment #6 from maksim at kde dot org  2010-06-04 00:55 -------
I can provide an .i file, but it would be on x86 (with 4.4.3). Is it likely to
be of use? 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[maksim at kde dot org]

------- Comment #7 from maksim at kde dot org  2010-06-04 16:38 -------
Created an attachment (id=20839)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20839&action=view)
Preprocessed source

Actually, never mind that --- got it on x86-64 from an another KDEer (credit to
Alex Fiestas).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[jamborm at gcc dot gnu dot org]

------- Comment #8 from jamborm at gcc dot gnu dot org  2010-06-07 16:56 -------
(In reply to comment #5)
> 
> have a look at PR44406... i believe its the same thing.
> 

I think it probably is because the patch of mine would lead to code
very similar to what exposed PR 44406.  However, PR 44406 actually no
longer happens on the trunk because it is very likely a duplicate of
PR 44164.

So can you please verify that this miscompilation still takes place
with the current 4.5 branch before I go into all the troubles of
looking at what compiles differently and where?  Thanks.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[pluto at agmk dot net]

------- Comment #9 from pluto at agmk dot net  2010-06-08 09:07 -------
(In reply to comment #8)
> (In reply to comment #5)
> > 
> > have a look at PR44406... i believe its the same thing.
> > 
> 
> I think it probably is because the patch of mine would lead to code
> very similar to what exposed PR 44406.  However, PR 44406 actually no
> longer happens on the trunk because it is very likely a duplicate of
> PR 44164.
> 
> So can you please verify that this miscompilation still takes place
> with the current 4.5 branch before I go into all the troubles of
> looking at what compiles differently and where?  Thanks.

verification in progress...


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[pluto at agmk dot net]

------- Comment #10 from pluto at agmk dot net  2010-06-08 10:15 -------
the gcc-4.5-20100608 still miscompiles kde4libs.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[jamborm at gcc dot gnu dot org]

------- Comment #11 from jamborm at gcc dot gnu dot org  2010-06-09 14:43 -------
OK, I have found the bug and I admit it is rather embarrassing.  I'll
submit a patch soon.


-- 

jamborm at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |jamborm at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2010-05-30 17:04:29         |2010-06-09 14:43:08
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5/4.6 Regression] possible SRA wrong-code generation.

[jamborm at gcc dot gnu dot org]

------- Comment #12 from jamborm at gcc dot gnu dot org  2010-06-10 16:49 -------
Subject: Bug 44258

Author: jamborm
Date: Thu Jun 10 16:49:09 2010
New Revision: 160561

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=160561
Log:
2010-06-10  Martin Jambor  <mjambor@suse.cz>

        PR tree-optimization/44258
        * tree-sra.c (build_access_subtree): Return false iff there is a
        partial overlap.
        (build_access_trees): Likewise.
        (analyze_all_variable_accesses): Disqualify candidates if
        build_access_trees returns true for them.

        * testsuite/gcc.dg/tree-ssa/pr44258.c: New test.


Added:
    trunk/gcc/testsuite/gcc.dg/tree-ssa/pr44258.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/tree-sra.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5 Regression] possible SRA wrong-code generation.

[pluto at agmk dot net]

------- Comment #13 from pluto at agmk dot net  2010-06-14 12:48 -------
do you plan to backport this fix to 4.5 branch?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5 Regression] possible SRA wrong-code generation.

[jamborm at gcc dot gnu dot org]

------- Comment #14 from jamborm at gcc dot gnu dot org  2010-06-14 12:51 -------
(In reply to comment #13)
> do you plan to backport this fix to 4.5 branch?
> 

Of course, I'm running the bootstrap and testsuite right now.  I will
commit it today if everything goes fine.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5 Regression] possible SRA wrong-code generation.

[jamborm at gcc dot gnu dot org]

------- Comment #15 from jamborm at gcc dot gnu dot org  2010-06-15 10:03 -------
Subject: Bug 44258

Author: jamborm
Date: Tue Jun 15 10:03:35 2010
New Revision: 160776

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=160776
Log:
2010-06-15  Martin Jambor  <mjambor@suse.cz>

        PR tree-optimization/44258
        * tree-sra.c (build_access_subtree): Return false iff there is a
        partial overlap.
        (build_access_trees): Likewise.
        (analyze_all_variable_accesses): Disqualify candidates if
        build_access_trees returns true for them.

        * testsuite/gcc.dg/tree-ssa/pr44258.c: New test.


Added:
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/tree-ssa/pr44258.c
Modified:
    branches/gcc-4_5-branch/gcc/ChangeLog
    branches/gcc-4_5-branch/gcc/testsuite/ChangeLog
    branches/gcc-4_5-branch/gcc/tree-sra.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258

[Bug tree-optimization/44258] [4.5 Regression] possible SRA wrong-code generation.

[jamborm at gcc dot gnu dot org]

------- Comment #16 from jamborm at gcc dot gnu dot org  2010-06-15 10:05 -------
This is now fixed on both the trunk and the 4.5 branch.


-- 

jamborm at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44258