[Bug ada/54688] New: [4.8 regression] a-ioexce.ads violation of implicit restriction "No_Elaboration_Code" breaks Ada bootstrap on sparc64-linux

[Bug ada/54688] New: [4.8 regression] a-ioexce.ads violation of implicit restriction "No_Elaboration_Code" breaks Ada bootstrap on sparc64-linux

[mikpe at it dot uu.se]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

             Bug #: 54688
           Summary: [4.8 regression] a-ioexce.ads violation of implicit
                    restriction "No_Elaboration_Code" breaks Ada bootstrap
                    on sparc64-linux
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: ada
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: mikpe@it.uu.se


Bootstrapping gcc-4.8-20120923 on sparc64-linux w/ Ada fails:

/mnt/scratch/objdir48/./prev-gcc/xgcc -B/mnt/scratch/objdir48/./prev-gcc/
-B/mnt/scratch/install48/sparc64-unknown-linux-gnu/bin/
-B/mnt/scratch/install48/sparc64-unknown-linux-gnu/bin/
-B/mnt/scratch/install48/sparc64-unknown-linux-gnu/lib/ -isystem
/mnt/scratch/install48/sparc64-unknown-linux-gnu/include -isystem
/mnt/scratch/install48/sparc64-unknown-linux-gnu/sys-include    -c -g -O2 
-gnatpg -gnata -W -Wall -nostdinc -I- -I. -Iada
-I/mnt/scratch/gcc-4.8-20120923/gcc/ada
-I/mnt/scratch/gcc-4.8-20120923/gcc/ada/gcc-interface
/mnt/scratch/gcc-4.8-20120923/gcc/ada/a-ioexce.ads -o ada/a-ioexce.o
a-ioexce.ads:21:19: violation of implicit restriction "No_Elaboration_Code"
a-ioexce.ads:22:19: violation of implicit restriction "No_Elaboration_Code"
a-ioexce.ads:23:19: violation of implicit restriction "No_Elaboration_Code"
a-ioexce.ads:24:19: violation of implicit restriction "No_Elaboration_Code"
a-ioexce.ads:25:19: violation of implicit restriction "No_Elaboration_Code"
a-ioexce.ads:26:19: violation of implicit restriction "No_Elaboration_Code"
a-ioexce.ads:27:19: violation of implicit restriction "No_Elaboration_Code"
a-ioexce.ads:28:19: violation of implicit restriction "No_Elaboration_Code"
make[3]: *** [ada/a-ioexce.o] Error 1
make[3]: *** Waiting for unfinished jobs....
make[3]: Leaving directory `/mnt/scratch/objdir48/gcc'
make[2]: *** [all-stage3-gcc] Error 2
make[2]: Leaving directory `/mnt/scratch/objdir48'
make[1]: *** [stage3-bubble] Error 2
make[1]: Leaving directory `/mnt/scratch/objdir48'
make: *** [bootstrap] Error 2

The previous weekly snapshot, gcc-4.8-20120916, did not have this problem.

Complete configuration parameters:
/mnt/scratch/gcc-4.8-20120923/configure --prefix=/mnt/scratch/install48
--with-gmp=/home/mikpe/pkgs/linux-sparc64/gmp-5.0.5
--with-mpfr=/home/mikpe/pkgs/linux-sparc64/mpfr-3.1.1
--with-mpc=/home/mikpe/pkgs/linux-sparc64/mpc-1.0.1 --with-cpu=v8
--enable-multilib --disable-plugin --disable-lto --disable-nls
--enable-threads=posix --enable-checking=release --disable-libmudflap
--enable-languages=c,c++,fortran,ada

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-09-24
                 CC|                            |ebotcazou at gcc dot
                   |                            |gnu.org
          Component|ada                         |bootstrap
   Target Milestone|---                         |4.8.0
            Summary|[4.8 regression]            |[4.8 regression] violation
                   |a-ioexce.ads violation of   |of implicit restriction
                   |implicit restriction        |"No_Elaboration_Code" on
                   |"No_Elaboration_Code"       |a-ioexce.ads
                   |breaks Ada bootstrap on     |
                   |sparc64-linux               |
     Ever Confirmed|0                           |1

--- Comment #1 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-24 19:04:29 UTC ---
Yep.  This is stage 3 so the stage 2 compiler is very likely miscompiled.

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
                 CC|ebotcazou at gcc dot        |
                   |gnu.org                     |
         AssignedTo|unassigned at gcc dot       |ebotcazou at gcc dot
                   |gnu.org                     |gnu.org

--- Comment #2 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-24 19:15:32 UTC ---
Investigating.

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bernds at gcc dot gnu.org

--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-26 11:43:40 UTC ---
It's r191493, the transformation is applied to a %sp-based load, which is
invalid since the frame is destroyed, and the resulting offset is bogus:

In .split4:

(insn 52 41 43 4 (set (reg:SI 1 %g1 [orig:137 D.6097+8 ] [137])
        (mem/c:SI (plus:SI (reg/f:SI 14 %sp)
                (const_int 80 [0x50])) [23 D.6097+8 S4 A64]))
/home/eric/svn/gcc/gcc/ada/restrict.adb:1067 61 {*movsi_insn}
     (expr_list:REG_EQUIV (mem/c:SI (plus:SI (reg/f:SI 14 %sp)
                (const_int 80 [0x50])) [23 D.6097+8 S4 A64])
        (nil)))

[...]
(note 60 43 61 4 NOTE_INSN_EPILOGUE_BEG)

(insn 61 60 62 4 (set (reg/f:SI 14 %sp)
        (minus:SI (reg/f:SI 14 %sp)
            (const_int -88 [0xffffffffffffffa8])))
/home/eric/svn/gcc/gcc/ada/restrict.adb:1068 183 {subsi3}

In .sched2:

(insn 61 64 52 4 (set (reg/f:SI 14 %sp)
        (minus:SI (reg/f:SI 14 %sp)
            (const_int -88 [0xffffffffffffffa8])))
/home/eric/svn/gcc/gcc/ada/restrict.adb:1068 183 {subsi3}
     (nil))

(insn 52 61 43 4 (set (reg:SI 1 %g1 [orig:137 D.6097+8 ] [137])
        (mem/c:SI (plus:SI (reg/f:SI 14 %sp)
                (const_int 168 [0xa8])) [23 D.6097+8 S4 A64]))
/home/eric/svn/gcc/gcc/ada/restrict.adb:1067 61 {*movsi_insn}
     (expr_list:REG_EQUIV (mem/c:SI (plus:SI (reg/f:SI 14 %sp)
                (const_int 80 [0x50])) [23 D.6097+8 S4 A64])
        (nil)))

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ro at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

Rainer Orth <ro at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|sparc64-linux               |sparc64-linux,
                   |                            |sparc*-*-solaris2*
                 CC|                            |ro at gcc dot gnu.org

--- Comment #4 from Rainer Orth <ro at gcc dot gnu.org> 2012-09-26 12:56:37 UTC ---
Same on Solaris/SPARC.

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[bernds at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #5 from Bernd Schmidt <bernds at gcc dot gnu.org> 2012-09-26 13:18:41 UTC ---
(In reply to comment #3)
> It's r191493, the transformation is applied to a %sp-based load, which is
> invalid since the frame is destroyed, and the resulting offset is bogus:
> 
> In .split4:
> 
> (insn 52 41 43 4 (set (reg:SI 1 %g1 [orig:137 D.6097+8 ] [137])
>         (mem/c:SI (plus:SI (reg/f:SI 14 %sp)
>                 (const_int 80 [0x50])) [23 D.6097+8 S4 A64]))
> /home/eric/svn/gcc/gcc/ada/restrict.adb:1067 61 {*movsi_insn}
>      (expr_list:REG_EQUIV (mem/c:SI (plus:SI (reg/f:SI 14 %sp)
>                 (const_int 80 [0x50])) [23 D.6097+8 S4 A64])
>         (nil)))
> 
> [...]
> (note 60 43 61 4 NOTE_INSN_EPILOGUE_BEG)
> 
> (insn 61 60 62 4 (set (reg/f:SI 14 %sp)
>         (minus:SI (reg/f:SI 14 %sp)
>             (const_int -88 [0xffffffffffffffa8])))
> /home/eric/svn/gcc/gcc/ada/restrict.adb:1068 183 {subsi3}
> 
> In .sched2:
> 
> (insn 61 64 52 4 (set (reg/f:SI 14 %sp)
>         (minus:SI (reg/f:SI 14 %sp)
>             (const_int -88 [0xffffffffffffffa8])))
> /home/eric/svn/gcc/gcc/ada/restrict.adb:1068 183 {subsi3}
>      (nil))
> 
> (insn 52 61 43 4 (set (reg:SI 1 %g1 [orig:137 D.6097+8 ] [137])
>         (mem/c:SI (plus:SI (reg/f:SI 14 %sp)
>                 (const_int 168 [0xa8])) [23 D.6097+8 S4 A64]))
> /home/eric/svn/gcc/gcc/ada/restrict.adb:1067 61 {*movsi_insn}
>      (expr_list:REG_EQUIV (mem/c:SI (plus:SI (reg/f:SI 14 %sp)
>                 (const_int 80 [0x50])) [23 D.6097+8 S4 A64])
>         (nil)))

I admit I don't quite see yet why this would be invalid (assuming that the
stack grows from high address to low ones). What do you mean by "the frame is
destroyed", is there something special about sparc?

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #6 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-26 13:37:25 UTC ---
> I admit I don't quite see yet why this would be invalid (assuming that the
> stack grows from high address to low ones). What do you mean by "the frame is
> destroyed", is there something special about sparc?

We go from

  g1 <- [sp+80]
  sp <- sp + 88

to

  sp <- sp + 88
  g1 <- [sp + 168]

so there is a double issue: the frame is destroyed after the %sp bump and the
new offset is bogus (it should be -8).  And the REG_EQUIV note isn't updated.

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[bernds at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #7 from Bernd Schmidt <bernds at gcc dot gnu.org> 2012-09-26 13:50:06 UTC ---
Oh, I missed that it was a MINUS rtx. This is noncanonical RTL, it should be
(plus (sp) (negated constant)).

Does the bug persist if you fix the sparc port in this way?

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #8 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-26 14:03:24 UTC ---
> Oh, I missed that it was a MINUS rtx. This is noncanonical RTL, it should be
> (plus (sp) (negated constant)).
> 
> Does the bug persist if you fix the sparc port in this way?

I'm not sure we want to change the SPARC port (this form is used because of the
larger offset range).  In any case, non-canonical RTL shouldn't result in wrong
code, only in missed optimization.

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[bernds at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #9 from Bernd Schmidt <bernds at gcc dot gnu.org> 2012-09-26 14:13:31 UTC ---
Created attachment 28283
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28283
Candidate patch.

The offset range should not be an issue with the right constraints etc.; the
port really ought to be changed.

In the meantime, please try this patch.

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ro at CeBiTec dot Uni-Bielefeld.DE]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #10 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> 2012-09-27 07:57:04 UTC ---
> --- Comment #9 from Bernd Schmidt <bernds at gcc dot gnu.org> 2012-09-26
> 14:13:31 UTC ---
> Created attachment 28283
>   --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28283
> Candidate patch.
[...]
> In the meantime, please try this patch.

Unfortunately, this doesn't help: same failure on sparc-sun-solaris2.1[01].

    Rainer

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[bernds at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #11 from Bernd Schmidt <bernds at gcc dot gnu.org> 2012-09-27 13:28:12 UTC ---
Hmm, strange. What if you just make a change like this:

@@ -4600,8 +4600,7 @@ parse_add_or_inc (struct mem_inc_info *m
   if (!REG_P (SET_DEST (pat)))
     return false;

-  if (GET_CODE (SET_SRC (pat)) != PLUS
-      && GET_CODE (SET_SRC (pat)) != MINUS)
+  if (GET_CODE (SET_SRC (pat)) != PLUS)
     return false;

   mii->inc_insn = insn;

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ro at CeBiTec dot Uni-Bielefeld.DE]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #12 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> 2012-09-27 15:59:38 UTC ---
> --- Comment #11 from Bernd Schmidt <bernds at gcc dot gnu.org> 2012-09-27 13:28:12 UTC ---
> Hmm, strange. What if you just make a change like this:
>
> @@ -4600,8 +4600,7 @@ parse_add_or_inc (struct mem_inc_info *m
>    if (!REG_P (SET_DEST (pat)))
>      return false;
>
> -  if (GET_CODE (SET_SRC (pat)) != PLUS
> -      && GET_CODE (SET_SRC (pat)) != MINUS)
> +  if (GET_CODE (SET_SRC (pat)) != PLUS)
>      return false;
>
>    mii->inc_insn = insn;

With that change, a sparc-sun-solaris2.11 bootstrap completed.  make
check is still running.

    Rainer

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #13 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-28 07:55:19 UTC ---
> The offset range should not be an issue with the right constraints etc.; the
> port really ought to be changed.

Yes, in fact we already have the machinery to do it.

> In the meantime, please try this patch.

Thanks, it works for me, modulo the obvious thinko (mii->inc_constant --> cst).

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[bernds at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

--- Comment #14 from Bernd Schmidt <bernds at gcc dot gnu.org> 2012-09-28 20:33:00 UTC ---
Author: bernds
Date: Fri Sep 28 20:32:55 2012
New Revision: 191838

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=191838
Log:
    PR bootstrap/54688
    * sched-deps.c (parse_add_or_inc): Remove MINUS handling.  Take
    STACK_GROWS_DOWNWARD into account.

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

[Bug bootstrap/54688] [4.8 regression] violation of implicit restriction "No_Elaboration_Code" on a-ioexce.ads

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54688

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

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

--- Comment #15 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-28 20:46:36 UTC ---
.