[Bug bootstrap/26679] New: boostrap failure due to warning in gcc/varasm.c

[Bug bootstrap/26679] New: boostrap failure due to warning in gcc/varasm.c

[gcc-bugzilla at gcc dot gnu dot org]

        During bootstrap line 3072 of varasm.c generates a warning that
        shift >= width of type, and the build dies due to -Werror.
        This diagnosis is correct but the shift is unreachable.

Environment:
System: Linux dps 2.6.15 #2 PREEMPT Sat Jan 7 17:47:27 GMT 2006 i686 GNU/Linux
Architecture: i686
host: i686-pc-linux-gnu
build: i686-pc-linux-gnu
target: i686-pc-linux-gnu
configured with: ../gcc/configure --prefix=/usr --enable-shared
--cache-file=config.cache

How-To-Repeat:
        bootstrap with HOST_WIDE_INT intendical to unsinged int (e.g.
        almost any 32 bit x86 box).


------- Comment #1 from dps at simpson dot demon dot co dot uk  2006-03-14 14:24 -------
Fix:
        This patch uses reprocessor logic to elinminate the (harmless,
        without -Werrror, warning. 
--- gcc/varasm.c.dist   2006-03-08 11:53:25.000000000 +0000
+++ gcc/varasm.c        2006-03-14 14:12:15.000000000 +0000
@@ -3067,11 +3067,16 @@
        int i;

        h ^= (hashval_t) hwi;
+#if HOST_BITS_PER_WIDE_INT != SIZEOF_UNSIGNED_INT * CHAR_BIT
+       /* Leaving this in emits a warning, so stop compilation, if n=1 */
        for (i = 1; i < n; ++i)
          {
            hwi >>= shift;
            h ^= (hashval_t) hwi;
          }
+#else
+       i=shift; i=n; /* Squash compiler warnings */
+#endif
       }
       break;

--- gcc/configure.ac.dist       2006-03-08 11:53:25.000000000 +0000
+++ gcc/configure.ac    2006-03-14 02:37:17.000000000 +0000
@@ -289,6 +289,7 @@
 AC_CHECK_SIZEOF(void *)
 AC_CHECK_SIZEOF(short)
 AC_CHECK_SIZEOF(int)
+AC_CHECK_SIZEOF(unsigned int)
 AC_CHECK_SIZEOF(long)
 AC_CHECK_TYPES([long long], [AC_CHECK_SIZEOF(long long)])
 AC_CHECK_TYPES([__int64], [AC_CHECK_SIZEOF(__int64)])
--- gcc/config.in.dist  2006-03-08 11:53:25.000000000 +0000
+++ gcc/config.in       2006-03-14 02:41:09.000000000 +0000
@@ -1240,6 +1240,11 @@
 #undef SIZEOF_INT
 #endif

+/* The size of a `usigned int` as computed by sizeof. */
+#ifndef USED_FOR_TARGET
+#undef SIZEOF_UNSIGNED_INT
+#endif
+

 /* The size of a `long', as computed by sizeof. */
 #ifndef USED_FOR_TARGET


-- 
           Summary: boostrap failure due to warning in gcc/varasm.c
           Product: gcc
           Version: 2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: dps at simpson dot demon dot co dot uk
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

[Bug bootstrap/26679] boostrap failure due to warning in gcc/varasm.c

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2006-03-14 14:29 -------
What compiler are you using to get that warning?
There should be no warning as shift is a variable and n is a variable and
should be zero.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|2.0                         |4.2.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

[Bug bootstrap/26679] boostrap failure due to warning in gcc/varasm.c

[pinskia at gcc dot gnu dot org]

------- Comment #3 from pinskia at gcc dot gnu dot org  2006-03-14 14:31 -------
And I don't see why you using SIZEOF_UNSIGNED_INT as unsigned int does not come
in anywhere.

Now hashval_t does but that could be anything.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

[Bug bootstrap/26679] boostrap failure due to warning in gcc/varasm.c

[joseph at codesourcery dot com]

------- Comment #4 from joseph at codesourcery dot com  2006-03-14 15:11 -------
Subject: Re:  boostrap failure due to warning in
 gcc/varasm.c

On Tue, 14 Mar 2006, pinskia at gcc dot gnu dot org wrote:

> What compiler are you using to get that warning?
> There should be no warning as shift is a variable and n is a variable and
> should be zero.

shift is a const variable initialized with a constant, so when building 
with optimization (this is the stage1 compiler building the stage2 
compiler) it gets replaced by its value.  Because the warning is given 
before dead code elimination, the fact that n is also a constant and the 
code is unreachable is irrelevant.  Why this error is newly appeared I'm 
not sure.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

[Bug bootstrap/26679] boostrap failure due to warning in gcc/varasm.c

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2006-03-14 15:24:03
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

[Bug middle-end/26679] [4.2 Regression] boostrap failure due to warning in gcc/varasm.c

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|bootstrap                   |middle-end
           Keywords|                            |build
            Summary|boostrap failure due to     |[4.2 Regression] boostrap
                   |warning in gcc/varasm.c     |failure due to warning in
                   |                            |gcc/varasm.c
   Target Milestone|---                         |4.2.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

[Bug middle-end/26679] [4.2 Regression] boostrap failure due to warning in gcc/varasm.c

[pinskia at gcc dot gnu dot org]

------- Comment #5 from pinskia at gcc dot gnu dot org  2006-03-14 15:27 -------
Actually can you try it again after:
2006-03-14  Richard Guenther  <rguenther@suse.de>

        * configure: Regenerate with autoconf 2.13.

I think the toplevel configure was exposing this.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

Re: [Bug bootstrap/26679] boostrap failure due to warning in gcc/varasm.c

[Graham Stott]

All,

If the warning isn't bogus then we probably need to do the shift in two steps
(i.e. hwi = (hwi >> (shift - 1)) >> 1) as done elsewhere to avoid the 
potential warning.

--- joseph at codesourcery dot com <gcc-bugzilla@gcc.gnu.org> wrote:

> 
> 
> ------- Comment #4 from joseph at codesourcery dot com  2006-03-14 15:11
> -------
> Subject: Re:  boostrap failure due to warning in
>  gcc/varasm.c
> 
> On Tue, 14 Mar 2006, pinskia at gcc dot gnu dot org wrote:
> 
> > What compiler are you using to get that warning?
> > There should be no warning as shift is a variable and n is a variable and
> > should be zero.
> 
> shift is a const variable initialized with a constant, so when building 
> with optimization (this is the stage1 compiler building the stage2 
> compiler) it gets replaced by its value.  Because the warning is given 
> before dead code elimination, the fact that n is also a constant and the 
> code is unreachable is irrelevant.  Why this error is newly appeared I'm 
> not sure.
> 
> 
> -- 
> 
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679
> 
> 

[Bug middle-end/26679] [4.2 Regression] boostrap failure due to warning in gcc/varasm.c

[graham dot stott at btinternet dot com]

------- Comment #6 from graham dot stott at btinternet dot com  2006-03-14 18:55 -------
Subject: Re:  boostrap failure due to warning in gcc/varasm.c

All,

If the warning isn't bogus then we probably need to do the shift in two steps
(i.e. hwi = (hwi >> (shift - 1)) >> 1) as done elsewhere to avoid the 
potential warning.

--- joseph at codesourcery dot com <gcc-bugzilla@gcc.gnu.org> wrote:

> 
> 
> ------- Comment #4 from joseph at codesourcery dot com  2006-03-14 15:11
> -------
> Subject: Re:  boostrap failure due to warning in
>  gcc/varasm.c
> 
> On Tue, 14 Mar 2006, pinskia at gcc dot gnu dot org wrote:
> 
> > What compiler are you using to get that warning?
> > There should be no warning as shift is a variable and n is a variable and
> > should be zero.
> 
> shift is a const variable initialized with a constant, so when building 
> with optimization (this is the stage1 compiler building the stage2 
> compiler) it gets replaced by its value.  Because the warning is given 
> before dead code elimination, the fact that n is also a constant and the 
> code is unreachable is irrelevant.  Why this error is newly appeared I'm 
> not sure.
> 
> 
> -- 
> 
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679
> 
> 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

[Bug middle-end/26679] [4.2 Regression] boostrap failure due to warning in gcc/varasm.c

[pinskia at physics dot uc dot edu]

------- Comment #7 from pinskia at gcc dot gnu dot org  2006-03-14 18:57 -------
Subject: Re:  boostrap failure due to warning in gcc/varasm.c


On Mar 14, 2006, at 1:55 PM, Graham Stott wrote:

> All,
>
> If the warning isn't bogus then we probably need to do the shift in 
> two steps
> (i.e. hwi = (hwi >> (shift - 1)) >> 1) as done elsewhere to avoid the
> potential warning.

The only reason why it is bogus is because well it is dead code :).


-- Pinski


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679

Re: [Bug bootstrap/26679] boostrap failure due to warning in gcc/varasm.c

[Andrew Pinski]

On Mar 14, 2006, at 1:55 PM, Graham Stott wrote:

> All,
>
> If the warning isn't bogus then we probably need to do the shift in 
> two steps
> (i.e. hwi = (hwi >> (shift - 1)) >> 1) as done elsewhere to avoid the
> potential warning.

The only reason why it is bogus is because well it is dead code :).


-- Pinski

[Bug middle-end/26679] [4.2 Regression] boostrap failure due to warning in gcc/varasm.c

[pinskia at gcc dot gnu dot org]

------- Comment #8 from pinskia at gcc dot gnu dot org  2006-03-15 13:13 -------
Fixed by:
2006-03-14  Richard Guenther  <rguenther@suse.de>

        * configure: Regenerate with autoconf 2.13.


-- 

pinskia at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26679