[Bug tree-optimization/59471] New: ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[Bug tree-optimization/59471] New: ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[jgreenhalgh at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

            Bug ID: 59471
           Summary: ICE using vector extensions (non-top-level
                    BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jgreenhalgh at gcc dot gnu.org

The following code:

typedef unsigned char uint8x8_t
  __attribute__ ((__vector_size__ (8)));

typedef unsigned short uint16x8_t
  __attribute__ ((__vector_size__ (16)));

typedef unsigned long uint64x2_t
  __attribute__ ((__vector_size__ (16)));

uint8x8_t
foo (uint16x8_t x)
{
  return (uint8x8_t) ((uint64x2_t) x)[0];
}

Will give this ICE for current trunk on AArch64, ARM and X86_64:

/work/build-x86/install/bin/gcc ../testcases/view-convert-expr.c -O3
../testcases/view-convert-expr.c: In function ‘foo’:
../testcases/view-convert-expr.c:11:1: error: non-top-level BIT_FIELD_REF,
IMAGPART_EXPR or REALPART_EXPR
 foo (uint16x8_t x)
 ^
BIT_FIELD_REF <VIEW_CONVERT_EXPR<vector(2) long unsigned int>(x), 64, 0>
../testcases/view-convert-expr.c:13:3: note: in statement
   return (uint8x8_t) ((uint64x2_t) x)[0];
   ^
D.1792 = VIEW_CONVERT_EXPR<uint8x8_t>(BIT_FIELD_REF
<VIEW_CONVERT_EXPR<vector(2) long unsigned int>(x), 64, 0>);
../testcases/view-convert-expr.c:11:1: internal compiler error: verify_gimple
failed
 foo (uint16x8_t x)
 ^
0x9b5a5a verify_gimple_in_cfg(function*)
    ../../src/gcc/gcc/tree-cfg.c:4837
0x8df347 execute_function_todo
    ../../src/gcc/gcc/passes.c:1847
0x8dfb73 execute_todo
    ../../src/gcc/gcc/passes.c:1877
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.

Looking at -fdump-tree-all-raw, I see this expression in
view-convert-expr.c.004t.gimple:

foo (uint16x8_t x)
gimple_bind <
  uint8x8_t D.1792;

  gimple_assign <view_convert_expr, D.1792,
VIEW_CONVERT_EXPR<uint8x8_t>(BIT_FIELD_REF <VIEW_CONVERT_EXPR<vector(2) long
unsigned int>(x), 64, 0>), NULL, NULL>
  gimple_return <D.1792 NULL>
>

For reference, my x86 compiler was configured as:

Configured with: ../src/gcc/configure --prefix=/work/build-x86/install
>From gcc-bugs-return-437333-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Wed Dec 11 15:40:38 2013
Return-Path: <gcc-bugs-return-437333-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 25649 invoked by alias); 11 Dec 2013 15:40:37 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 25595 invoked by uid 48); 11 Dec 2013 15:40:30 -0000
From: "janus at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug fortran/59440] [4.9 Regression] ICE in force_decl_die, at dwarf2out.c:20111 with -g
Date: Wed, 11 Dec 2013 15:40:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: fortran
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords: ice-on-valid-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: janus at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.9.0
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: cc
Message-ID: <bug-59440-4-qtn5uldq0T@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59440-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59440-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2013-12/txt/msg00988.txt.bz2
Content-length: 434

http://gcc.gnu.org/bugzilla/show_bug.cgi?idY440

janus at gcc dot gnu.org changed:

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

--- Comment #1 from janus at gcc dot gnu.org ---
(Maybe-not-so-) Wild guess: The regression is most probably due to r205679 (cf
PR 37132).

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-12-13
          Component|tree-optimization           |middle-end
            Version|unknown                     |4.9.0
   Target Milestone|---                         |4.9.0
            Summary|ICE using vector extensions |[4.9 Regression] ICE using
                   |(non-top-level              |vector extensions
                   |BIT_FIELD_REF,              |(non-top-level
                   |IMAGPART_EXPR or            |BIT_FIELD_REF,
                   |REALPART_EXPR)              |IMAGPART_EXPR or
                   |                            |REALPART_EXPR)
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I think BIT_FIELD_REF's type can't be a vector, so it has to be integral type
in this case.

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

--- Comment #5 from Marc Glisse <glisse at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #4)
> I think BIT_FIELD_REF's type can't be a vector,

Er, I am quite sure a BIT_FIELD_REF can be a vector. Maybe that wasn't a
general statement and I missed the context?

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
You mean BIT_FIELD_REF argument can be a vector?  Sure.  But the type of the
BIT_FIELD_REF itself?

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

--- Comment #7 from Marc Glisse <glisse at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #6)
> You mean BIT_FIELD_REF argument can be a vector?  Sure.  But the type of the
> BIT_FIELD_REF itself?

Yes, the type of the BIT_FIELD_REF itself. A quick grep gives:
                    t = build3 (BIT_FIELD_REF, vectype, new_temp,
in tree-vect-stmts.c where I assume vectype is a vector type. IIRC,
tree-vect-generic.c also produces plenty of those when lowering extra-long
vectors.

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org

--- Comment #9 from Richard Biener <rguenth at gcc dot gnu.org> ---
If nobody beats me to it I'll take care of the gimplify.c piece after holidays.

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

--- Comment #8 from Richard Biener <rguenth at gcc dot gnu.org> ---
The reason I added this verification is that code does not expect those to
appear at non-outermost handled-component.  a V_C_E around a vector type
B_F_R should have been split into two different stmts with a temporary.
Eventually gimplification doesn't ensure this (and thus force_gimple_operand
and friends may generate invalid gimple).

So the fix probably has a piece in gimplify.c at least - make sure
is_gimple_reg_type () B_F_R, RP_E, IP_E get a temporary.

RP_E and IP_E are probably safe as complex types are quite constrained.

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> ---
Testing

Index: gimplify.c                                                               
===================================================================             
--- gimplify.c  (revision 205891)                                               
+++ gimplify.c  (working copy)                                                  
@@ -7358,12 +7358,22 @@ gimplify_expr (tree *expr_p, gimple_seq                 
                                        TREE_TYPE (*expr_p));                   
          break;                                                                

+       case VIEW_CONVERT_EXPR:                                                 
+         if (is_gimple_reg_type (TREE_TYPE (*expr_p))                          
+             && is_gimple_reg_type (TREE_TYPE (TREE_OPERAND (*expr_p, 0))))    
+           {                                                                   
+             ret = gimplify_expr (&TREE_OPERAND (*expr_p, 0), pre_p,           
+                                  post_p, is_gimple_val, fb_rvalue);           
+             recalculate_side_effects (*expr_p);                               
+             break;                                                            
+           }                                                                   
+         /* Fallthru.  */                                                      
+                                                                               
        case ARRAY_REF:                                                         
        case ARRAY_RANGE_REF:                                                   
        case REALPART_EXPR:                                                     
        case IMAGPART_EXPR:                                                     
        case COMPONENT_REF:                                                     
-       case VIEW_CONVERT_EXPR:                                                 
          ret = gimplify_compound_lval (expr_p, pre_p, post_p,                  
                                        fallback ? fallback : fb_rvalue);       
          break;                                                                
@@ -7709,10 +7719,17 @@ gimplify_expr (tree *expr_p, gimple_seq                 
          break;                                                                

        case BIT_FIELD_REF:                                                     
-         ret = gimplify_expr (&TREE_OPERAND (*expr_p, 0), pre_p,               
-                              post_p, is_gimple_lvalue, fb_either);            
-         recalculate_side_effects (*expr_p);                                   
-         break;                                                                
+         {                                                                     
+           if (is_gimple_reg_type (TREE_TYPE (*expr_p))                        
+               && is_gimple_reg_type (TREE_TYPE (TREE_OPERAND (*expr_p, 0))))  
+             ret = gimplify_expr (&TREE_OPERAND (*expr_p, 0), pre_p,           
+                                  post_p, is_gimple_val, fb_rvalue);           
+           else                                                                
+             ret = gimplify_expr (&TREE_OPERAND (*expr_p, 0), pre_p,           
+                                  post_p, is_gimple_lvalue, fb_either);        
+           recalculate_side_effects (*expr_p);                                 
+           break;                                                              
+         }                                                                     

        case TARGET_MEM_REF:                                                    
          {                                                      


code-generation for the testcase is awkward though:

foo:
.LFB0:  
        .cfi_startproc
        movaps  %xmm0, -24(%rsp)
        movq    -24(%rsp), %rax
        movq    %rax, -24(%rsp)
        movq    -24(%rsp), %xmm0
        ret

not sure why RTL cannot recover here, we expand from

foo (uint16x8_t x)
{
  vector(2) long unsigned int _2;
  long unsigned int _3;
  uint8x8_t _4;

  <bb 2>:
  _2 = VIEW_CONVERT_EXPR<vector(2) long unsigned int>(x_5(D));
  _3 = BIT_FIELD_REF <_2, 64, 0>;
  _4 = VIEW_CONVERT_EXPR<uint8x8_t>(_3);
  return _4;

with forwprop we could re-write this into

  _4 = BIT_FIELD_REF <x_5(D), 64, 0>;

as BIT_FIELD_REF is allowed to select uint8x8_t from uint16x8_t directly.

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

--- Comment #11 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Wed Jan  8 08:59:29 2014
New Revision: 206420

URL: http://gcc.gnu.org/viewcvs?rev=206420&root=gcc&view=rev
Log:
2014-01-08  Richard Biener  <rguenther@suse.de>

    PR middle-end/59471
    * gimplify.c (gimplify_expr): Gimplify register-register type
    VIEW_CONVERT_EXPRs to separate stmts.

    * gcc.dg/pr59471.c: New testcase.

Added:
    trunk/gcc/testsuite/gcc.dg/pr59471.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/gimplify.c
    trunk/gcc/testsuite/ChangeLog

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

Richard Biener <rguenth at gcc dot gnu.org> changed:

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

--- Comment #12 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.

[Bug middle-end/59471] [4.9 Regression] ICE using vector extensions (non-top-level BIT_FIELD_REF, IMAGPART_EXPR or REALPART_EXPR)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59471

--- Comment #13 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Author: jakub
Date: Wed Jan  8 22:33:12 2014
New Revision: 206448

URL: http://gcc.gnu.org/viewcvs?rev=206448&root=gcc&view=rev
Log:
    PR middle-end/59471
    * gcc.dg/pr59471.c (foo): Avoid vector type arguments or return
    type, use pointers to vector type instead.

Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/pr59471.c