[Bug c/16925] New: ICC when building a m68hc11 cross-compiler on 64-bit architectures

[Bug c/16925] New: ICC when building a m68hc11 cross-compiler on 64-bit architectures

[aurelien at aurel32 dot net]

I am unable to build a m68hc11 cross-compiler on 64-bit architectures. xgcc
generates an internal compiler error during the build of libgcc2:

make[1]: Entering directory `/home/aurel32/gcc/gcc-3.4.1/gcc'
for d in libgcc m68hc12 libgcc/m68hc12 mshort libgcc/mshort fshort-double
libgcc/fshort-double mshort/fshort-double libgcc/mshort/fshort-double
m68hc12/mshort libgcc/m68hc12/mshort m68hc12/fshort-double
libgcc/m68hc12/fshort-double m68hc12/mshort/fshort-double
libgcc/m68hc12/mshort/fshort-double; do \
  if [ -d $d ]; then true; else /bin/sh ./mkinstalldirs $d; fi; \
done
if [ -f stmp-dirs ]; then true; else touch stmp-dirs; fi
./xgcc -B./ -B/home/aurel32/m68hc11/bin/ -isystem /home/aurel32/m68hc11/include
-isystem /home/aurel32/m68hc11/sys-include
-L/home/aurel32/gcc/gcc-3.4.1/gcc/../ld -Os -mrelax  -DUSE_GAS -DIN_GCC -g 
-DIN_LIBGCC2 -Dinhibit_libc -I. -I. -I. -I./. -I./../include 
-DFINE_GRAINED_LIBRARIES -DL_pack_df -c dp-bit.c -o libgcc/./_pack_df.o
dp-bit.c: In function `__pack_d':
dp-bit.c:434: error: could not split insn
(insn:HI 61 613 63 2 dp-bit.c:219 (parallel [
            (set (mem:DI (plus:HI (reg:HI 2 y)
                        (const_int 19 [0x13])) [11 fraction+0 S8 A8])
                (ior:DI (mem:DI (plus:HI (reg:HI 2 y)
                            (const_int 19 [0x13])) [11 fraction+0 S8 A8])
                    (const_int 2251799813685248 [0x8000000000000])))
            (clobber (reg:HI 1 d))
        ]) 81 {iordi3} (nil)
    (expr_list:REG_DEAD (reg:HI 2 y)
        (expr_list:REG_UNUSED (reg:HI 1 d)
            (expr_list:REG_UNUSED (reg:HI 1 d)
                (nil)))))
dp-bit.c:434: internal compiler error: in final_scan_insn, at final.c:2429
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.
make[1]: *** [libgcc/./_pack_df.o] Error 1
make[1]: Leaving directory `/home/aurel32/gcc/gcc-3.4.1/gcc'


The problem is _not_ present in gcc 3.2.3, but present in gcc 3.3.1, gcc 3.3.2,
gcc 3.4.0 and gcc 3.4.1

The problem seems to be present only on 64-bit architectures (alpha, amd64 and
ia64), and in all cases, the problem occurs on the same file, same line, with
the same error message.

configure line used:
./configure --prefix=/home/aurel32 --with-gnu-as --with-gnu-ld --enable-nls
--without-included-gettext --disable-checking --enable-languages=c,c++
--target=m68hc11 --without-headers

-- 
           Summary: ICC when building a m68hc11 cross-compiler on 64-bit
                    architectures
           Product: gcc
           Version: 3.4.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: aurelien at aurel32 dot net
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: alphapca56-unknown-linux-gnu
  GCC host triplet: alphapca56-unknown-linux-gnu
GCC target triplet: m68hc11


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug c/16925] ICC when building a m68hc11 cross-compiler on 64-bit architectures

[aurelien at aurel32 dot net]

------- Additional Comments From aurelien at aurel32 dot net  2004-08-08 23:24 -------
Created an attachment (id=6907)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=6907&action=view)
preprocessed file


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICC when building a m68hc11 cross-compiler on 64-bit architectures

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |target
           Keywords|                            |build, ice-on-valid-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICC when building a m68hc11 cross-compiler on 64-bit architectures

[aurelien at aurel32 dot net]

------- Additional Comments From aurelien at aurel32 dot net  2004-08-11 22:25 -------
Reduced testcase:

typedef unsigned int UDItype __attribute__ ((mode (DI)));

void ICE_on_64bit_archs (void)
{
  UDItype a;
  a |= 0x8000000000000LL;
}

It compiles well on 32-bit machines, and I get an ICE on 64-bit machines:

test2.c: In function `ICE_on_64bit_archs':
test2.c:7: error: could not split insn
(insn 9 17 11 (parallel [
            (set (mem/f:DI (plus:HI (reg:HI 0 x)
                        (const_int 1 [0x1])) [0 a+0 S8 A64])
                (ior:DI (mem/f:DI (plus:HI (reg:HI 0 x)
                            (const_int 1 [0x1])) [0 a+0 S8 A64])
                    (const_int 2251799813685248 [0x8000000000000])))
            (clobber (reg:HI 1 d))
        ]) 81 {iordi3} (nil)
    (nil))
test2.c:7: internal compiler error: in final_scan_insn, at final.c:2429
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICC when building a m68hc11 cross-compiler on 64-bit architectures

[aurelien at aurel32 dot net]

------- Additional Comments From aurelien at aurel32 dot net  2004-09-11 13:36 -------
Created an attachment (id=7096)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7096&action=view)
Patch to fix that bug

The first part of the patch adds the fact that CONST_INT could be a DImode rtx,
that needs to be split into two SImode rtx.

The second part fixes the core of the bug. On 32-bit hosts, this part is called
when a 32-bit integer has two be split into two 32-bit integers (CONST_INT is a
32-bit integer on 32-bit hosts, as well as SI mode). The problem is that
CONST_INT is a 64-bit integer on 64-bit hosts.

Actually this condition is always false on 32-bit hosts (a grep for
split_logical shows that the operands is always a DImode when mode == SImode).
This should also be always false on 64-bit hosts, that's why I removed that
code. 

I don't understand why that code is trying to convert one SImode rtx into two
SImode rtx, moreover in a function whose name is beginning by split.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[schwab at suse dot de]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|ICC when building a m68hc11 |ICE when building a m68hc11
                   |cross-compiler on 64-bit    |cross-compiler on 64-bit
                   |architectures               |architectures


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[dbaird at nmt dot edu]

------- Additional Comments From dbaird at nmt dot edu  2004-12-22 07:13 -------
Just for the record, I am having the same problem.  I am using GCC as
the host compiler.  My CHOST is x86_64-pc-linux-gnu.

I am building gcc-3.3.4 with patches from (Pre-Release 2.92):
http://m68hc11.serveftp.org/m68hc1x-builder-2.92.tar.gz

Compilation succeeds after I apply gcc-3.4.2-16925.patch from above.

(In reply to comment #3)
> Created an attachment (id=7096)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7096&action=view)
> Patch to fix that bug
> 
> The first part of the patch adds the fact that CONST_INT could be a DImode rtx,
> that needs to be split into two SImode rtx.
> 
> The second part fixes the core of the bug. On 32-bit hosts, this part is called
> when a 32-bit integer has two be split into two 32-bit integers (CONST_INT is a
> 32-bit integer on 32-bit hosts, as well as SI mode). The problem is that
> CONST_INT is a 64-bit integer on 64-bit hosts.
> 
> Actually this condition is always false on 32-bit hosts (a grep for
> split_logical shows that the operands is always a DImode when mode == SImode).
> This should also be always false on 64-bit hosts, that's why I removed that
> code. 
> 
> I don't understand why that code is trying to convert one SImode rtx into two
> SImode rtx, moreover in a function whose name is beginning by split.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
  GCC build triplet|alphapca56-unknown-linux-gnu|
   GCC host triplet|alphapca56-unknown-linux-gnu|Any 64bit target (or really
                   |                            |LP64)
   Last reconfirmed|0000-00-00 00:00:00         |2004-12-22 07:20:41
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[aurelien at aurel32 dot net]

------- Additional Comments From aurelien at aurel32 dot net  2004-12-22 09:15 -------
           What    |Removed                     |Added
----------------------------------------------------------------------------
  GCC build triplet|alphapca56-unknown-linux-gnu|
   GCC host triplet|alphapca56-unknown-linux-gnu|Any 64bit target (or really
                   |                            |LP64)

I think it should rather be "Any 64bit host", as the target is "m68hc11"

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[ciceron at gcc dot gnu dot org]

------- Additional Comments From ciceron at gcc dot gnu dot org  2005-01-28 20:34 -------
Mine. 

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-05-08 10:09 -------
Subject: Bug 16925

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_4-branch
Changes by:	ciceron@gcc.gnu.org	2005-05-08 10:08:50

Modified files:
	gcc            : ChangeLog 
	gcc/config/m68hc11: m68hc11.c 

Log message:
	PR target/16925
	* config/m68hc11/m68hc11.c (m68hc11_gen_highpart): Handle split of
	64-bit constants on 64-bit hosts.
	(m68hc11_split_logical): Simplify.
	(m68hc11_split_move): Likewise.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=2.2326.2.854&r2=2.2326.2.855
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/m68hc11/m68hc11.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.91.6.6&r2=1.91.6.7



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-05-08 10:14 -------
Subject: Bug 16925

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-4_0-branch
Changes by:	ciceron@gcc.gnu.org	2005-05-08 10:14:31

Modified files:
	gcc            : ChangeLog 
	gcc/config/m68hc11: m68hc11.c 

Log message:
	PR target/16925
	* config/m68hc11/m68hc11.c (m68hc11_gen_highpart): Handle split of
	64-bit constants on 64-bit hosts.
	(m68hc11_split_logical): Simplify.
	(m68hc11_split_move): Likewise.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.7592.2.227&r2=2.7592.2.228
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/m68hc11/m68hc11.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=1.111&r2=1.111.8.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-05-08 10:23 -------
Subject: Bug 16925

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	ciceron@gcc.gnu.org	2005-05-08 10:23:40

Modified files:
	gcc            : ChangeLog 
	gcc/config/m68hc11: m68hc11.c 

Log message:
	PR target/16925
	* config/m68hc11/m68hc11.c (m68hc11_gen_highpart): Handle split of
	64-bit constants on 64-bit hosts.
	(m68hc11_split_logical): Simplify.
	(m68hc11_split_move): Likewise.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.8655&r2=2.8656
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/m68hc11/m68hc11.c.diff?cvsroot=gcc&r1=1.114&r2=1.115



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[ciceron at gcc dot gnu dot org]

------- Additional Comments From ciceron at gcc dot gnu dot org  2005-05-08 10:29 -------
Fixed in 3_4 (3.4.4?), 4_0 (4.0.1), and mainline.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925

[Bug target/16925] ICE when building a m68hc11 cross-compiler on 64-bit architectures

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |3.4.4 4.0.1 4.1.0
   Target Milestone|---                         |3.4.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16925