[Bug target/15289] New: reload error with big-endian subregs

[Bug target/15289] New: reload error with big-endian subregs

[amodra at bigpond dot net dot au]

/* On powerpc-linux -O1, this testcase ICEs on hammer branch, and
   generates wrong code on gcc-3.4.0 and mainline.  */

typedef struct { _Complex char a; _Complex char b; } Scc2;

Scc2 s = { 1+2i, 3+4i };

int checkScc2 (Scc2 s)
{
  return s.a != 1+2i || s.b != 3+4i;
}

int main (void)
{
  return checkScc2 (s);
}

-- 
           Summary: reload error with big-endian subregs
           Product: gcc
           Version: 3.5.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P2
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: amodra at bigpond dot net dot au
                CC: gcc-bugs at gcc dot gnu dot org
GCC target triplet: powerpc-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug target/15289] reload error with big-endian subregs

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-05 06:30 -------
Confirmed, also happens on powerpc-darwin.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
 GCC target triplet|powerpc-linux               |powerpc-linux, powerpc-
                   |                            |darwin
   Last reconfirmed|0000-00-00 00:00:00         |2004-05-05 06:30:26
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug target/15289] reload error with big-endian subregs

[amodra at bigpond dot net dot au]

------- Additional Comments From amodra at bigpond dot net dot au  2004-05-05 14:24 -------
This:
(insn 16 15 17 0 (set (reg:SI 9 9 [129])
        (zero_extend:SI (subreg:QI (reg:SI 0 0 [128]) 2))) 18 {*rs6000.md:508}
(insn_list 13 (nil))
    (nil))

gets turned into
(insn 64 15 16 0 (set (reg:QI 10 10)
        (reg:QI 0 0)) -1 (nil)
    (nil))

(insn 16 64 17 0 (set (reg:SI 9 9 [129])
        (zero_extend:SI (reg:QI 10 10))) 18 {*rs6000.md:508} (insn_list 13 (nil))
    (nil))

Jan, http://gcc.gnu.org/ml/gcc-patches/2003-04/msg00882.html added the code to
reload that made a reload necessary for (subreg:QI (reg:SI 0 0 [128]) 2) with
the hunk
*************** find_reloads (insn, replace, ind_levels,
*** 2869,2874 ****
--- 2876,2887 ----
  	      if (GET_CODE (SUBREG_REG (operand)) == REG
  		  && REGNO (SUBREG_REG (operand)) < FIRST_PSEUDO_REGISTER)
  		{
+ 		  if (!subreg_offset_representable_p
+ 			(REGNO (SUBREG_REG (operand)),
+ 			 GET_MODE (SUBREG_REG (operand)),
+ 			 SUBREG_BYTE (operand),
+ 			 GET_MODE (operand)))
+ 		     force_reload = 1;
  		  offset += subreg_regno_offset (REGNO (SUBREG_REG (operand)),
  						 GET_MODE (SUBREG_REG (operand)),
  						 SUBREG_BYTE (operand),


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] reload error with non-lowpart subregs

[amodra at bigpond dot net dot au]

------- Additional Comments From amodra at bigpond dot net dot au  2004-05-27 13:36 -------
Also happens on x86

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|target                      |rtl-optimization
 GCC target triplet|powerpc-linux, powerpc-     |powerpc-linux, powerpc-
                   |darwin                      |darwin, i686-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/3.5 Regression] reload error with non-lowpart subregs

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
OtherBugsDependingO|16028                       |
              nThis|                            |
            Summary|reload error with non-      |[3.4/3.5 Regression] reload
                   |lowpart subregs             |error with non-lowpart
                   |                            |subregs
   Target Milestone|---                         |3.4.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/3.5 Regression] reload error with non-lowpart subregs

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-08-19 20:53 -------
Jan, can you explain the code you added to reload?  It looks like there should
be a comment on that code, if nothing else.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jh at suse dot cz


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/3.5 Regression] reload error with non-lowpart subregs

[hubicka at gcc dot gnu dot org]

------- Additional Comments From hubicka at gcc dot gnu dot org  2004-08-26 12:18 -------
As explained in the cited email, the idea of patch is to avoid silent
misscompilations in the cases similar to those.  Here we have single 32bit
register holding 2 8bit fields and we happen to generate SUBREG:QI accessing one
placed in the middle of the word.

i386 has hardware support for this, but I doubt PPC does and we used to simply
pass the subreg around pass reload and forget about the funny byte offset and
finally produe read of the incorrect word.

The particular hunk of code looks quite self explaining - if the subreg can not
be represented by the hardware, we must do something.  I can add this comment to
the code if it seems benefical.

Reload is supposed to know how to reload all subregs that are produced earlier,
if he does not, it is either bug to produce one or bug in reload dealing with
this particular case.  I will try to look into how the pre 3.4 GCC happent to
deal with this.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/3.5 Regression] reload error with non-lowpart subregs

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-08-29 19:02 -------
Postponed until GCC 3.4.3.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.2                       |3.4.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-10-31 02:07 -------
Jan, this is OK for mainline and *3.4.4*, if tests still pass.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.3                       |3.4.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2004-11-28 13:08 -------
Can someone reconfirm this bug?  Does this still happen with the subreg 
validation patches? 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[amodra at bigpond dot net dot au]

------- Additional Comments From amodra at bigpond dot net dot au  2004-11-28 13:27 -------
/src/tmp/pr15289.c: In function 'checkScc2':
/src/tmp/pr15289.c:9: internal compiler error: in gen_highpart, at emit-rtl.c:1260


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-11-28 17:54 -------
RTH, this is another subreg quirk. Care to take a look?

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[rth at gcc dot gnu dot org]

------- Additional Comments From rth at gcc dot gnu dot org  2004-11-29 21:24 -------
Sigh.  Mine.  I hate gen_imagpart.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |rth at gcc dot gnu dot org
                   |dot org                     |
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-01 18:14 -------
Subject: Bug 15289

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	rth@gcc.gnu.org	2004-12-01 18:13:36

Modified files:
	gcc            : ChangeLog emit-rtl.c expmed.c expr.c function.c 
	                 rtl.h 

Log message:
	PR rtl-opt/15289
	* emit-rtl.c (gen_complex_constant_part): Remove.
	(gen_realpart, gen_imagpart): Remove.
	* rtl.h (gen_realpart, gen_imagpart): Remove.
	* expmed.c (extract_bit_field): Remove CONCAT hack catering to
	gen_realpart/gen_imagpart.
	* expr.c (write_complex_part, read_complex_part): New.
	(emit_move_via_alt_mode, emit_move_via_integer, emit_move_resolve_push,
	emit_move_complex_push, emit_move_complex, emit_move_ccmode,
	emit_move_multi_word): Split out from ...
	(emit_move_insn_1): ... here.
	(expand_expr_real_1) <COMPLEX_EXPR>: Use write_complex_part.
	<REALPART_EXPR, IMAGPART_EXPR>: Use read_complex_part.
	* function.c (assign_parm_setup_reg): Hard-code transformations
	instead of using gen_realpart/gen_imagpart.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.6662&r2=2.6663
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/emit-rtl.c.diff?cvsroot=gcc&r1=1.429&r2=1.430
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/expmed.c.diff?cvsroot=gcc&r1=1.210&r2=1.211
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/expr.c.diff?cvsroot=gcc&r1=1.749&r2=1.750
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/function.c.diff?cvsroot=gcc&r1=1.591&r2=1.592
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/rtl.h.diff?cvsroot=gcc&r1=1.528&r2=1.529



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-01 18:16 -------
Subject: Bug 15289

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	rth@gcc.gnu.org	2004-12-01 18:15:59

Added files:
	gcc/testsuite/gcc.c-torture/execute: 20041201-1.c 

Log message:
	Test for PR 15289.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/execute/20041201-1.c.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-02 19:16 -------
Subject: Bug 15289

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_4-branch
Changes by:	rth@gcc.gnu.org	2004-12-02 19:16:38

Modified files:
	gcc            : ChangeLog emit-rtl.c expr.c function.c 
	                 integrate.c optabs.c rtl.h 

Log message:
	PR rtl-opt/15289
	* emit-rtl.c (gen_complex_constant_part): Remove.
	(gen_realpart, gen_imagpart, subreg_realpart_p): Remove.
	* expr.c (write_complex_part, read_complex_part): New.
	(emit_move_via_alt_mode, emit_move_via_integer, emit_move_resolve_push,
	emit_move_complex_push, emit_move_complex, emit_move_ccmode,
	emit_move_multi_word): Split out from ...
	(emit_move_insn_1): ... here.
	(expand_expr_real) <COMPLEX_EXPR>: Use write_complex_part.
	<REALPART_EXPR, IMAGPART_EXPR>: Use read_complex_part.
	<CONJ_EXPR>: Likewise.
	* function.c (assign_parms): Hard-code transformations
	instead of using gen_realpart/gen_imagpart.
	* integrate.c (initialize_for_inline): Likewise.
	* optabs.c (expand_unop): Use read_complex_part/write_complex_part.
	(expand_complex_abs): Likewise.
	(expand_binop): Likewise.  Rearrange to build a CONCAT at the end,
	rather than creating a complex target at the beginning.
	* rtl.h (gen_realpart, gen_imagpart, subreg_realpart_p): Remove.
	(read_complex_part, write_complex_part): Declare.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=2.2326.2.715&r2=2.2326.2.716
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/emit-rtl.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.365.4.6&r2=1.365.4.7
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/expr.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.615.4.15&r2=1.615.4.16
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/function.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.483.4.19&r2=1.483.4.20
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/integrate.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.244.2.1&r2=1.244.2.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/optabs.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.204.4.2&r2=1.204.4.3
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/rtl.h.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.448.4.7&r2=1.448.4.8



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[rth at gcc dot gnu dot org]

------- Additional Comments From rth at gcc dot gnu dot org  2004-12-03 23:26 -------
Fixed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|rth at gcc dot gnu dot org  |
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-05 05:21 -------
Subject: Bug 15289

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_4-branch
Changes by:	rth@gcc.gnu.org	2004-12-05 05:21:02

Modified files:
	gcc            : ChangeLog emit-rtl.c expr.c function.c 
	                 integrate.c optabs.c rtl.h 

Log message:
	* emit-rtl.c, expr.c, function.c, integrate.c, optabs.c, rtl.h:
	Revert the patches for PR rtl-opt/15289.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=2.2326.2.719&r2=2.2326.2.720
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/emit-rtl.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.365.4.7&r2=1.365.4.8
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/expr.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.615.4.17&r2=1.615.4.18
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/function.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.483.4.20&r2=1.483.4.21
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/integrate.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.244.2.3&r2=1.244.2.4
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/optabs.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.204.4.3&r2=1.204.4.4
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/rtl.h.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.448.4.8&r2=1.448.4.9



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[rth at gcc dot gnu dot org]

------- Additional Comments From rth at gcc dot gnu dot org  2004-12-05 05:22 -------
Won't fix for 3.4.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.4                       |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289

[Bug rtl-optimization/15289] [3.4/4.0 Regression] reload error with non-lowpart subregs

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-12-05 05:38 -------
*** Bug 16028 has been marked as a duplicate of this bug. ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fjahanian at apple dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15289