[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[laurent at guerby dot net]

------- Comment #1 from laurent at guerby dot net  2006-01-03 19:44 -------
The Ada runtime does not use C complex primitives. In this case, the culprit in
all three case is likely to be the following code in ada/a-ngelfu.adb

   ------------
   -- Arctan --
   ------------

   --  Natural cycle

   function Arctan
     (Y    : Float_Type'Base;
      X    : Float_Type'Base := 1.0)
      return Float_Type'Base
   is
   begin
      if X = 0.0
        and then Y = 0.0
      then
         raise Argument_Error;

      elsif Y = 0.0 then
         if X > 0.0 then
            return 0.0;
         else -- X < 0.0
            return Pi * Float_Type'Copy_Sign (1.0, Y);
         end if;

      elsif X = 0.0 then
         if Y > 0.0 then
            return Half_Pi;
         else -- Y < 0.0
            return -Half_Pi;
         end if;

      else
         return Local_Atan (Y, X);
      end if;
   end Arctan;

In all three failing case (log, cos,, we end up calling Log(-1.0) which calls
Arctan (Y => 0.0, X => -1.0) which ends up in the

return Pi * Float_Type'Copy_Sign (1.0, Y);

case, so I assume 'Copy_Sign is broken in this case, Dave could you try the
following small program with your hppa-linux 4.0.x GCC:

guerby@m:~/tmp/pr20754> cat > p.adb
with ada.text_io; use ada.text_io;
procedure p is
begin
   put_line(long_float'copy_sign(1.0,0.0)'img);
end p;
guerby@m:~/tmp/pr20754> gnatmake p
gcc -c p.adb
gnatbind -x p.ali
gnatlink p.ali
guerby@m:~/tmp/pr20754> ./p
 1.00000000000000E+00

Laurent


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

------- Comment #2 from danglin at gcc dot gnu dot org  2006-01-03 20:27 -------
dave@hiauly6:~/gcc_test$ ./p
 1.00000000000000E+00


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

------- Comment #3 from danglin at gcc dot gnu dot org  2006-01-03 20:29 -------
That was with 4.0.0.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[laurent at guerby dot net]

------- Comment #4 from laurent at guerby dot net  2006-01-03 21:19 -------
May be this one? 

guerby@m:~/tmp/pr20754> cat > p2.adb
with Ada.Numerics.Elementary_Functions;
with Ada.Numerics.Complex_Types; 
with Ada.Numerics.Complex_Elementary_Functions; 
with Ada.Text_IO;

procedure P2 is

   use Ada.Numerics.Elementary_Functions;
   use Ada.Numerics.Complex_Types;
   use Ada.Numerics.Complex_Elementary_Functions;
   use Ada.Text_IO;

   procedure Put (C : in Complex) is
   begin
      Put_Line ("(" & Float'Image (Re (C)) & ", "
                & Float'Image (Im (C)) & ")");
   end Put;

  C0 : constant Complex := (-1.0, 0.0);
  C1 : Constant Complex := Log (C0);

begin
   Put (C0);
   Put (C1);
   Put_Line (Float'Image (Arctan (Y => 0.0, X => -1.0)));
end P2;
guerby@m:~/tmp/pr20754> gnatmake p2
gcc -c p2.adb
gnatbind -x p2.ali
gnatlink p2.ali
guerby@m:~/tmp/pr20754> ./p2
(-1.00000E+00,  0.00000E+00)
( 0.00000E+00,  3.14159E+00)
 3.14159E+00


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[dave at hiauly1 dot hia dot nrc dot ca]

------- Comment #5 from dave at hiauly1 dot hia dot nrc dot ca  2006-01-04 02:41 -------
Subject: Re:  ACATS cxg1005 fails at runtime on hppa-linux

> May be this one? 

Nope.

There seems to be an optimisation issue.  I rebuilt the runtime
at -O0.  The test passes when compiled at -O0 and -O1, but fails at
-O2.

Dave


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[laurent at guerby dot net]

------- Comment #6 from laurent at guerby dot net  2006-01-04 11:48 -------
(In reply to comment #5)
> There seems to be an optimisation issue.  I rebuilt the runtime
> at -O0.  The test passes when compiled at -O0 and -O1, but fails at
> -O2.

Hmm less likely to be fixed on 4.0 then. Could you try cxg1005 with 4.1 and 4.2
compiler? Your testresults runs appear to end before reaching this test.

Laurent


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

------- Comment #7 from danglin at gcc dot gnu dot org  2006-01-05 02:53 -------
The bug is as follows.  The return from the call to
cxg1005__test_block__complex_pack__compose_from_cartesian___389
is in registers ret0 and ret1.

        bl cxg1005__test_block__complex_pack__compose_from_cartesian___389,%r2
        fcpy,sgl %fr12L,%fr4L
        stw %r28,-16(%r30)
        fldws -16(%r30),%fr19L
        copy %r0,%r19
        copy %r0,%r20
        .loc 4 88 0
        fcpy,sgl %fr13L,%fr4L
        stw %r19,-496(%r30)
        stw %r20,-492(%r30)
        stw %r19,-488(%r30)
        stw %r20,-484(%r30)
        .loc 4 87 0
        .loc 4 88 0
        bl
cxg1005__test_block__complex_pack__compose_from_cartesian__3___395,%r
2
        stw %r29,-16(%r30)
        .loc 4 87 0
        fldws -16(%r30),%fr21L

Prior to the call to
cxg1005__test_block__complex_pack__compose_from_cartesian__3___395, the
return value in ret1 is saved in the frame marker at -16(%r30).  The
function cxg1005__test_block__complex_pack__compose_from_cartesian__3___395
is a leaf function and doesn't allocate a frame.  It does a similar
register interchange clobbering the old value at -16(%r30).

        .type  
cxg1005__test_block__complex_pack__compose_from_cartesian__3___3
95, @function
.LFB50:
        .loc 1 474 0
cxg1005__test_block__complex_pack__compose_from_cartesian__3___395:
        .PROC
        .CALLINFO FRAME=0,NO_CALLS
        .ENTRY
.LVL3:
.LVL4:
        .loc 1 477 0
        fstws %fr4L,-16(%r30)
        copy %r0,%r28
        bv %r0(%r2)
        ldw -16(%r30),%r29
        .EXIT
        .PROCEND


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[laurent at guerby dot net]

------- Comment #8 from laurent at guerby dot net  2006-01-05 08:07 -------
This is with 4.0, right?

Are you able to produce a C99 or Fortran test case from your analsys?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug target/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

------- Comment #9 from danglin at gcc dot gnu dot org  2006-01-07 18:37 -------
I'm changing this to a target bug.  The bug is in the define for
SECONDARY_MEMORY_NEEDED_RTX(MODE):

#define SECONDARY_MEMORY_NEEDED_RTX(MODE) \
  gen_rtx_MEM (MODE, gen_rtx_PLUS (Pmode, stack_pointer_rtx, GEN_INT (-16)))

There are a number of places where this memory location gets used during
final code generation.  I don't think we can add clobbers.

When we define SECONDARY_MEMORY_NEEDED_RTX as above, reload introduces
store-load insns into the RTL to copy a value from a general/floating
register to a floating/general register.  Subsequently, scheduling may
copy separate the load and store insns.  This then can introduce a conflict
with the use of this memory location during code generation.

I'm not sure about the validity of scheduling moving the "fldws
-16(%r30),%fr21L"
over a call.  The call_insn is marked as a const or pure call:

(call_insn/u 80 4722 4769 0 cxg1005.adb:88 (parallel [
            (set (reg:SC 28 %r28)
                (call (mem:SI (symbol_ref/v:SI
("@cxg1005__test_block__complex_p
ack__compose_from_cartesian__3___395") <function_decl 0x40100620
cxg1005__test_b
lock__complex_pack__compose_from_cartesian__3>) [0 S4 A32])
                    (const_int 16 [0x10])))
            (clobber (reg:SI 1 %r1))
            (clobber (reg:SI 2 %r2))
            (use (const_int 0 [0x0]))
        ]) 204 {call_val_symref} (insn_list:REG_DEP_TRUE 60
(insn_list:REG_DEP_O
UTPUT 63 (insn_list:REG_DEP_ANTI 66 (insn_list:REG_DEP_ANTI 65
(insn_list:REG_DE
P_OUTPUT 20 (insn_list:REG_DEP_ANTI 22 (insn_list:REG_DEP_OUTPUT 21
(insn_list:R
EG_DEP_TRUE 79 (insn_list:REG_DEP_ANTI 64 (nil))))))))))
    (expr_list:REG_DEAD (reg:SF 32 %fr4)
        (expr_list:REG_UNUSED (reg:SI 2 %r2)
            (expr_list:REG_UNUSED (reg:SI 1 %r1)
                (nil))))
    (expr_list:REG_DEP_TRUE (use (reg:SF 32 %fr4))
        (nil)))

The decision as to whether the function is pure is made before reload and
that doesn't change when the SECONDARY_MEMORY_NEEDED_RTX MEM is used (ie.,
we use a special location in the frame of the caller for the copy.  This
problem can be avoided if we always ensure that the function has a frame.
However, this adds instructions and still doesn't fix the issue that the
MEM might be used between the store and load in other situations.

Thus, I think the best fix is to remove the SECONDARY_MEMORY_NEEDED_RTX
define and do the register copy in the move patterns directly.  This keeps
the MEM from being exposed in the RTL.  We lose the performance benefit
of scheduling the store and load.  However, we don't need a frame to
do floating-general copies.

As an aside, the problem is introduced in the RTL in the sched2 pass:

call_insn/u 80 4722 5740 0 cxg1005.adb:88 (parallel [
            (set (reg:SC 28 %r28)
                (call (mem:SI (symbol_ref/v:SI
("@cxg1005__test_block__complex_p
ack__compose_from_cartesian__3___395") <function_decl 0x40100620
cxg1005__test_b
lock__complex_pack__compose_from_cartesian__3>) [0 S4 A32])
                    (const_int 16 [0x10])))
            (clobber (reg:SI 1 %r1))
            (clobber (reg:SI 2 %r2))
            (use (const_int 0 [0x0]))
        ]) 204 {call_val_symref} (insn_list:REG_DEP_OUTPUT 63
(insn_list:REG_DEP
_ANTI 4765 (insn_list:REG_DEP_ANTI 4763 (insn_list:REG_DEP_OUTPUT 12
(insn_list:
REG_DEP_OUTPUT 10 (insn_list:REG_DEP_OUTPUT 17 (insn_list:REG_DEP_ANTI 22
(insn_
list:REG_DEP_OUTPUT 15 (insn_list:REG_DEP_ANTI 4767 (insn_list:REG_DEP_ANTI
4768
 (insn_list:REG_DEP_OUTPUT 4677 (insn_list:REG_DEP_ANTI 5372
(insn_list:REG_DEP_
ANTI 5391 (insn_list:REG_DEP_ANTI 5392 (insn_list:REG_DEP_ANTI 5393
(insn_list:R
EG_DEP_ANTI 5394 (insn_list:REG_DEP_ANTI 5395 (insn_list:REG_DEP_ANTI 5396
(insn
_list:REG_DEP_ANTI 5397 (insn_list:REG_DEP_ANTI 5398 (insn_list:REG_DEP_ANTI
539
9 (insn_list:REG_DEP_ANTI 5400 (insn_list:REG_DEP_OUTPUT 5390
(insn_list:REG_DEP
_TRUE 79 (insn_list:REG_DEP_TRUE 5373 (insn_list:REG_DEP_ANTI 64
(nil)))))))))))
))))))))))))))))
    (expr_list:REG_DEAD (reg:SF 32 %fr4)
        (expr_list:REG_UNUSED (reg:SI 2 %r2)
            (expr_list:REG_UNUSED (reg:SI 1 %r1)
                (nil))))
    (expr_list:REG_DEP_TRUE (use (reg:SF 32 %fr4))
        (nil)))

(note 5740 80 4766 0 ("cxg1005.adb") 87)

(insn 4766 5740 5741 0 cxg1005.adb:87 (set (reg:SF 66 %fr21 [orig:778+4 ]
[778])
        (mem:SF (plus:SI (reg/f:SI 30 %r30)
                (const_int -16 [0xfffffff0])) [0 S4 A32])) 77 {*pa.md:4325}
(ins
n_list:REG_DEP_ANTI 5391 (insn_list:REG_DEP_TRUE 5373 (insn_list:REG_DEP_ANTI
64
 (insn_list:REG_DEP_TRUE 4763 (insn_list:REG_DEP_TRUE 4765 (nil))))))
    (nil))



-- 

danglin at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|ada                         |target


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug target/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

------- Comment #10 from danglin at gcc dot gnu dot org  2006-01-11 00:07 -------
Subject: Bug 20754

Author: danglin
Date: Wed Jan 11 00:07:16 2006
New Revision: 109557

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109557
Log:
        PR target/20754
        * pa.md: Create separate 32 and 64-bit move patterns for SI, DI, SF
        and DF modes.  Add alternatives to copy between general and floating
        point registers to the 32-bit patterns.
        * pa-64.h (SECONDARY_MEMORY_NEEDED_RTX): Delete undefine.
        * pa.h (SECONDARY_MEMORY_NEEDED_RTX): Delete define.
        (SECONDARY_MEMORY_NEEDED): Secondary memory is only needed when
        generating 64-bit code.
        * pa.c (output_move_double): Handle copies between general and
        floating registers.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/pa/pa-64.h
    trunk/gcc/config/pa/pa.c
    trunk/gcc/config/pa/pa.h
    trunk/gcc/config/pa/pa.md


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug target/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

------- Comment #11 from danglin at gcc dot gnu dot org  2006-01-11 00:28 -------
Subject: Bug 20754

Author: danglin
Date: Wed Jan 11 00:28:13 2006
New Revision: 109566

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109566
Log:
        PR target/20754
        * pa.md: Create separate 32 and 64-bit move patterns for SI, DI, SF
        and DF modes.  Add alternatives to copy between general and floating
        point registers to the 32-bit patterns.
        * pa-64.h (SECONDARY_MEMORY_NEEDED_RTX): Delete undefine.
        * pa.h (SECONDARY_MEMORY_NEEDED_RTX): Delete define.
        (SECONDARY_MEMORY_NEEDED): Secondary memory is only needed when
        generating 64-bit code.
        * pa.c (output_move_double): Handle copies between general and
        floating registers.


Modified:
    branches/gcc-4_1-branch/gcc/ChangeLog
    branches/gcc-4_1-branch/gcc/config/pa/pa-64.h
    branches/gcc-4_1-branch/gcc/config/pa/pa.c
    branches/gcc-4_1-branch/gcc/config/pa/pa.h
    branches/gcc-4_1-branch/gcc/config/pa/pa.md


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug target/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

------- Comment #12 from danglin at gcc dot gnu dot org  2006-01-11 00:44 -------
Subject: Bug 20754

Author: danglin
Date: Wed Jan 11 00:43:56 2006
New Revision: 109568

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109568
Log:
        PR target/20754
        * pa.md: Create separate 32 and 64-bit move patterns for SI, DI, SF
        and DF modes.  Add alternatives to copy between general and floating
        point registers to the 32-bit patterns.
        * pa-64.h (SECONDARY_MEMORY_NEEDED_RTX): Delete undefine.
        * pa.h (SECONDARY_MEMORY_NEEDED_RTX): Delete define.
        (SECONDARY_MEMORY_NEEDED): Secondary memory is only needed when
        generating 64-bit code.
        * pa.c (output_move_double): Handle copies between general and
        floating registers.


Modified:
    branches/gcc-4_0-branch/gcc/ChangeLog
    branches/gcc-4_0-branch/gcc/config/pa/pa-64.h
    branches/gcc-4_0-branch/gcc/config/pa/pa.c
    branches/gcc-4_0-branch/gcc/config/pa/pa.h
    branches/gcc-4_0-branch/gcc/config/pa/pa.md


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug target/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

------- Comment #13 from danglin at gcc dot gnu dot org  2006-01-18 16:34 -------
Fixed by patches on 4.0, 4.1 and trunk.


-- 

danglin at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug target/20754] ACATS cxg1005 fails at runtime on hppa-linux

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.0.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754

[Bug ada/20754] ACATS cxg1005 fails at runtime on hppa-linux

[danglin at gcc dot gnu dot org]

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20754