[Bug target/37270] New: [4.4 Regression] SH: spill failure for R0_REGS

[Bug target/37270] New: [4.4 Regression] SH: spill failure for R0_REGS

[kkojima at gcc dot gnu dot org]

The trunk compiler fails to compile libffi.call/cls_8byte.c
with -O2 -fomit-frame-pointer.  Here is a reduced test case:

typedef struct s8
{
  int a;
  float b;
} s8;

s8
s8_fn (struct s8 a1, struct s8 a2)
{
  struct s8 result;

  result.b = a1.b + a2.b;
  bar (a1.a, a1.b, a2.a, a2.b, result.a, result.b);
  return result;
}

With -O2 -fomit-frame-pointer, it causes

foo.c:15: error: unable to find a register to spill in class 'R0_REGS'
foo.c:15: error: this is the insn:
(insn:HI 23 44 45 3 foo.c:14 (parallel [
            (set (subreg:SF (reg:SI 1 r1 [orig:174 D.1226+4 ] [174]) 0)
                (reg:SF 163 [ D.1218 ]))
            (use (reg/v:PSI 151 ))
            (clobber (scratch:SI))
        ]) 205 {movsf_ie} (expr_list:REG_DEAD (reg:SF 163 [ D.1218 ])
        (nil)))
foo.c:15: internal compiler error: in spill_failure, at reload1.c:2098


-- 
           Summary: [4.4 Regression] SH: spill failure for R0_REGS
           Product: gcc
           Version: 4.4.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: kkojima at gcc dot gnu dot org
GCC target triplet: sh4-unknown-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37270

[Bug target/37270] [4.4 Regression] SH: spill failure for R0_REGS

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2008-08-29 04:45 -------
Do you know when this started to happen?  I almost want to say this is related
to IRA.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
           Keywords|                            |ra
   Target Milestone|---                         |4.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37270

[Bug target/37270] [4.4 Regression] SH: spill failure for R0_REGS

[kkojima at gcc dot gnu dot org]

------- Comment #2 from kkojima at gcc dot gnu dot org  2008-08-29 05:00 -------
(In reply to comment #1)

It starts to fail at some point between revision 139589 and 139690.
I guess that the IRA merge reveals the latent issue of the sh backend.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37270

[Bug target/37270] [4.4 Regression] SH: spill failure for R0_REGS

[kkojima at gcc dot gnu dot org]

------- Comment #3 from kkojima at gcc dot gnu dot org  2008-08-30 22:24 -------
Subject: Bug 37270

Author: kkojima
Date: Sat Aug 30 22:23:20 2008
New Revision: 139809

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=139809
Log:
        PR target/37270
        * config/sh/sh.h (LEGITIMIZE_RELOAD_ADDRESS): Generate
        the reload address with the index register for SFmode
        access with a displacement.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/sh/sh.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37270

[Bug target/37270] [4.4 Regression] SH: spill failure for R0_REGS

[kkojima at gcc dot gnu dot org]

------- Comment #4 from kkojima at gcc dot gnu dot org  2008-08-31 00:28 -------
Fixed.


-- 

kkojima at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37270