[Bug c/12640] New: SEGV building tk

[Bug c/12640] New: SEGV building tk

[green at redhat dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640

           Summary: SEGV building tk
           Product: gcc
           Version: tree-ssa
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: green at redhat dot com
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu

I will upload the problematic .i file.

$ gcc --version
gcc (GCC) 3.5-tree-ssa 20031015 (merged 20031005)
Copyright (C) 2003 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

$ gcc -O1 tkUnixSend.i -S
gcc: Internal error: Segmentation fault (program cc1)
Please submit a full bug report.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

[Bug c/12640] SEGV building tk

[green at redhat dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640



------- Additional Comments From green at redhat dot com  2003-10-16 14:48 -------
Created an attachment (id=4942)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=4942&action=view)
Preprocessed input source to reproduce bug.

Compile with -O1 to reproduce failure.

[Bug c/12640] SEGV building tk

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640



------- Additional Comments From pinskia at gcc dot gnu dot org  2003-10-16 21:32 -------
What happens if you make the stacklimit higher?
I think there is already a fixme in one of the files for something like this.

[Bug optimization/12640] stack overflow building tk

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |optimization
           Keywords|                            |ice-on-valid-code
            Summary|SEGV building tk            |stack overflow building tk
   Target Milestone|---                         |tree-ssa


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-10-17 13:38 -------
On powerpc-apple-darwin (where is a catcher for stack overflow in gcc):
[il0503d-dhcp76:~/src/gccPR] apinski% ~/gccssa_cbased/bin/gcc pr12640.i -O1
Out of stack space.
Try running `limit stacksize unlimited' in the shell to raise its limit.
Here is the top n lines of the backtrace:
#4  0x001c6de0 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1875
#5  0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#6  0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#7  0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#8  0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#9  0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#10 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#11 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#12 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#13 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#14 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#15 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#16 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#17 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#18 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#19 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#20 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#21 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#22 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#23 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#24 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#25 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917
#26 0x001c6e48 in get_strlen (arg=0x19400100) at /Users/apinski/src/gccssa/src/gcc/tree-ssa-
ccp.c:1917

[Bug optimization/12640] stack overflow building tk

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dnovillo at gcc dot gnu dot
                   |                            |org


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-10-17 13:40 -------
Most likely caused by:
2003-07-31  Diego Novillo  <dnovillo@redhat.com>
        
        * Makefile.in (tree-ssa-ccp.o): Depend on $(EXPR_H).
        * builtins.c (c_strlen): Remove static declaration.
        (simplify_builtin_fputs): Remove static declaration.
        (simplify_builtin_sprintf): New local function.
        (expand_builtin_sprintf): Remove by surrounding with #if 0.
        (expand_builtin): Add BUILT_IN_SPRINTF to the list of built-ins
        handed over to simplify_builtin.
        (validate_arglist): Do not allow arguments with TREE_SIDE_EFFECTS.
        (simplify_builtin_fputs): Add new argument KNOWN_LEN.  If it's set,
        use it instead of trying to compute the length of the string.
        Update all callers.
        * expr.h (simplify_builtin_fputs): Declare.
        * tree-flow.h (fold_stmt): Change argument type to tree *.  Update
        all users.
        * tree-ssa-ccp.c: Include expr.h.
        (replace_uses_in): If the statement makes a call to some selected
        built-ins, mark it for folding.
        (get_strlen): New local function.
        (ccp_fold_builtin): New local function. 
        (fold_stmt): Call it.
        (set_rhs): Fix if-else-if chaining.  Handle cases where the whole
        statement needs to be replaced.
        * tree.h (c_strlen): Declare.

[Bug optimization/12640] [tree-ssa] stack overflow building tk

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|stack overflow building tk  |[tree-ssa] stack overflow
                   |                            |building tk

[Bug optimization/12640] [tree-ssa] stack overflow building tk

[green at redhat dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640



------- Additional Comments From green at redhat dot com  2003-11-02 02:29 -------
The work around is to build with -fno-builtin.

[Bug optimization/12640] [tree-ssa] strlen builtin causes stack overflow building tk

[green at redhat dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640


green at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[tree-ssa] stack overflow   |[tree-ssa] strlen builtin
                   |building tk                 |causes stack overflow
                   |                            |building tk

[Bug optimization/12640] [tree-ssa] strlen builtin causes stack overflow building tk

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2003-11-12 17:57 -------
Just an infinite loop in tree-ssa-ccp.c:get_strlen() 
 
Smaller test case: 
 
typedef struct PendingCommand 
{ 
  char *result; 
  struct PendingCommand *nextPtr; 
} PendingCommand; 
 
static void 
SendEventProc (char *resultString) 
{ 
  char *p; 
  PendingCommand *pc, *pcPtr; 
 
  resultString = ""; 
  while (*p == '-') 
    { 
      if (p[2] == ' ') 
        { 
          resultString = p + 3; 
        } 
    } 
  for (pcPtr = pc; pcPtr != ((void *) 0); pcPtr = pcPtr->nextPtr) 
    { 
      pcPtr->result = malloc (strlen (resultString) + 1); 
      strcpy (pcPtr->result, resultString); 
    } 
} 
 
Diego, your patch/function cause this, perhaps you can have a look? 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2003-11-12 17:57:58
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640

[Bug optimization/12640] [tree-ssa] strlen builtin causes stack overflow building tk

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2003-11-12 18:22 -------
Small, smaller, smallest... 
 
int i; 
static void 
SendEventProc (char *resultString) 
{ 
  char *p; 
 
  resultString = ""; 
  while (*p == '-') 
    { 
      if (p[2] == ' ') 
        { 
          resultString = p + 3; 
        } 
    } 
  for (;;) 
    { 
      i = strlen (resultString) + 1; 
    } 
} 
 
The thing is that we have this PHI: 
resultString_1 = PHI <(char *)""(0), resultString_1(1), resultString_9(2)>; 
 
and we go on to try and get the lentgh of all the PHI args, by looking at the 
DEF stmts of each PHI arg... and resultString_1 has itself as an argument.  
There's your infinite loop. 
 
Probably need a patch like this (perhaps just return NULL_TREE, but I don't 
think we have to...)  No time to test or think, so I've just put it here so 
that it doesn't get lost in my messy file system. 
 
ndex: tree-ssa-ccp.c 
=================================================================== 
RCS file: /cvs/gcc/gcc/gcc/Attic/tree-ssa-ccp.c,v 
retrieving revision 1.1.2.108 
diff -c -3 -p -r1.1.2.108 tree-ssa-ccp.c 
*** tree-ssa-ccp.c      11 Nov 2003 18:04:46 -0000      1.1.2.108 
--- tree-ssa-ccp.c      12 Nov 2003 18:20:25 -0000 
*************** get_strlen (tree arg) 
*** 1922,1928 **** 
          arg_len = prev_arg_len = NULL_TREE; 
          for (i = 0; i < PHI_NUM_ARGS (def_stmt); i++) 
            { 
!             arg_len = get_strlen (PHI_ARG_DEF (def_stmt, i)); 
              if (arg_len == NULL_TREE) 
                return NULL_TREE; 
 
--- 1922,1939 ---- 
          arg_len = prev_arg_len = NULL_TREE; 
          for (i = 0; i < PHI_NUM_ARGS (def_stmt); i++) 
            { 
!             tree arg_def_stmt = PHI_ARG_DEF (def_stmt, i); 
! 
!             /* If this PHI has itself as an argument, we cannot 
!                determine the string length of this argument.  However, 
!                if we can find a constant string length for the other 
!                PHI args then we can still be sure that this is a 
!                constant string length.  So be optimistic and just 
!                continue with the next argument.  */ 
!             if (arg_def_stmt == def_stmt) 
!               continue; 
! 
!             arg_len = get_strlen (arg_def_stmt); 
              if (arg_len == NULL_TREE) 
                return NULL_TREE; 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640

[Bug optimization/12640] [tree-ssa] strlen builtin causes stack overflow building tk

[dnovillo at redhat dot com]

------- Additional Comments From dnovillo at redhat dot com  2003-11-12 18:32 -------
Subject: Re:  [tree-ssa] strlen builtin causes
	stack overflow building tk

On Wed, 2003-11-12 at 13:22, steven at gcc dot gnu dot org wrote:

> ndex: tree-ssa-ccp.c 
> =================================================================== 
> RCS file: /cvs/gcc/gcc/gcc/Attic/tree-ssa-ccp.c,v 
> retrieving revision 1.1.2.108 
> diff -c -3 -p -r1.1.2.108 tree-ssa-ccp.c 
> *** tree-ssa-ccp.c      11 Nov 2003 18:04:46 -0000      1.1.2.108 
> --- tree-ssa-ccp.c      12 Nov 2003 18:20:25 -0000 
> *************** get_strlen (tree arg) 
> *** 1922,1928 **** 
>           arg_len = prev_arg_len = NULL_TREE; 
>           for (i = 0; i < PHI_NUM_ARGS (def_stmt); i++) 
>             { 
> !             arg_len = get_strlen (PHI_ARG_DEF (def_stmt, i)); 
>               if (arg_len == NULL_TREE) 
>                 return NULL_TREE; 
>  
> --- 1922,1939 ---- 
>           arg_len = prev_arg_len = NULL_TREE; 
>           for (i = 0; i < PHI_NUM_ARGS (def_stmt); i++) 
>             { 
> !             tree arg_def_stmt = PHI_ARG_DEF (def_stmt, i); 
> ! 
> !             /* If this PHI has itself as an argument, we cannot 
> !                determine the string length of this argument.  However, 
> !                if we can find a constant string length for the other 
> !                PHI args then we can still be sure that this is a 
> !                constant string length.  So be optimistic and just 
> !                continue with the next argument.  */ 
> !             if (arg_def_stmt == def_stmt) 
> !               continue; 
> ! 
> !             arg_len = get_strlen (arg_def_stmt); 
>               if (arg_len == NULL_TREE) 
>                 return NULL_TREE; 
>  
Looks correct.  Care to take it through a full bootstrap/test cycle?


Diego.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640

[Bug optimization/12640] [tree-ssa] strlen builtin causes stack overflow building tk

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-13 00:41 -------
Patch was submitted: <http://gcc.gnu.org/ml/gcc-patches/2003-11/msg00900.html>.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640

[Bug optimization/12640] [tree-ssa] strlen builtin causes stack overflow building tk

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2003-11-13 01:04 -------
Subject: Bug 12640

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	tree-ssa-20020619-branch
Changes by:	steven@gcc.gnu.org	2003-11-13 01:04:19

Modified files:
	gcc            : ChangeLog.tree-ssa tree-ssa-ccp.c 

Log message:
	PR optimization/12640
	* tree-ssa-ccp.c (get_strlen): Don't follow the UD chain of a PHI argument if
	the DEF stmt for the argument is the PHI itself.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.tree-ssa.diff?cvsroot=gcc&only_with_tag=tree-ssa-20020619-branch&r1=1.1.2.852&r2=1.1.2.853
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-ssa-ccp.c.diff?cvsroot=gcc&only_with_tag=tree-ssa-20020619-branch&r1=1.1.2.108&r2=1.1.2.109



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640

[Bug optimization/12640] [tree-ssa] strlen builtin causes stack overflow building tk

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-13 01:22 -------
Fixed.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12640