bootstrap broken on powerpc-linux and powerpc64-linux

[Alan Modra <amodra at bigpond dot net dot au>]

Both powerpc-linux and powerpc64-linux die compiling libgfortran.  Looks
to be your recent patches.  The powerpc64 failure is

matmul_c4.c: In function 'matmul_c4':
matmul_c4.c:274: internal compiler error: in rtx_equiv_p, at struct-equiv.c:644

(gdb) up
#2  0x00000000103bb790 in set_dest_equiv_p (x=0x8000679010, y=0x80006791d0, 
    info=0x1ffffffeb80) at /src/gcc-current/gcc/struct-equiv.c:761
(gdb) p debug_rtx (x)
(parallel [
        (set (parallel:SC [
                    (expr_list:REG_DEP_ANTI (reg:SF 33 1)
                        (const_int 0 [0x0]))
                    (expr_list:REG_DEP_ANTI (reg:SF 34 2)
                        (const_int 4 [0x4]))
                ])
            (call (mem:SI (symbol_ref:DI ("__mulsc3") [flags 0x41] <function_decl 0x8000299200 __mulsc3>) [0 S4 A8])
                (const_int 64 [0x40])))
        (use (const_int 0 [0x0]))
        (clobber (reg:SI 65 lr))
    ])

That one looks relatively easy to fix.

The powerpc-linux one is
/home/alan/build/ppc/gcc-curr/./gcc/xgcc
-B/home/alan/build/ppc/gcc-curr/./gcc/ -B/usr/local/powerpc-linux/bin/
-B/usr/local/powerpc-linux/lib/ -isystem /usr/local/powerpc-linux/include
-isystem /usr/local/powerpc-linux/sys-include -DHAVE_CONFIG_H
-I. -I/src/gcc-current/libgfortran -I. -iquote/src/gcc-current/libgfortran/io
-I/src/gcc-current/libgfortran/../gcc
-I/src/gcc-current/libgfortran/../gcc/config -I../.././gcc -D_GNU_SOURCE
-std=gnu99 -Wall -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition -Wextra -Wwrite-strings -O2 -g -O2 -c
/src/gcc-current/libgfortran/generated/product_c8.c  -fPIC -DPIC -o
.libs/product_c8.o
product_c8.c: In function 'product_c8':
product_c8.c:173: internal compiler error: in note_local_live, at struct-equiv.c:315

(gdb) up
#1  0x10385628 in note_local_live (info=0xffffdc88, x=0x40247fa8, 
    y=<value optimized out>, rvalue=<value optimized out>)
    at /src/gcc-current/gcc/struct-equiv.c:315
(gdb) 
#2  0x10386068 in rtx_equiv_p (xp=<value optimized out>, y=0x40247c90, 
    rvalue=1, info=0xffffdc88) at /src/gcc-current/gcc/struct-equiv.c:538
(gdb) p debug_rtx (x)
(reg:DF 5 5 [orig:272+8 ] [272])
$2 = void
(gdb) p debug_rtx (y)
(reg:DF 45 13 [266])
$3 = void
(gdb) up
#3  0x10385a24 in rtx_equiv_p (xp=0x402369ac, y=0x4024a3d0, rvalue=-1, 
    info=0xffffdc88) at /src/gcc-current/gcc/struct-equiv.c:557
(gdb) 
#4  0x10386b54 in insns_match_p (i1=0x40236990, i2=0x40243cf0, 
    info=0xffffdc88) at /src/gcc-current/gcc/struct-equiv.c:926
(gdb) p debug_rtx (i1)
(insn:HI 341 340 342 32 /src/gcc-current/libgfortran/generated/product_c8.c:141 (set (mem/s:DF (plus:SI (reg/v/f:SI 28 28 [orig:163 dest ] [163])
                (const_int 8 [0x8])) [17 S8 A64])
        (reg:DF 5 5 [orig:272+8 ] [272])) 328 {*movdf_hardfloat32} (nil)
    (expr_list:REG_DEAD (reg:DF 5 5 [orig:272+8 ] [272])
        (nil)))
$4 = void
(gdb) p debug_rtx (i2)
(insn:HI 301 299 603 29 /src/gcc-current/libgfortran/generated/product_c8.c:133 (set (mem:DF (plus:SI (reg/v/f:SI 28 28 [orig:163 dest ] [163])
                (const_int 8 [0x8])) [17 S8 A64])
        (reg:DF 45 13 [266])) 328 {*movdf_hardfloat32} (insn_list:REG_DEP_TRUE 300 (nil))
    (expr_list:REG_DEAD (reg:DF 45 13 [266])
        (nil)))
$5 = void

This may not be enough info for you to figure out what is going on,
because I see rtx_equiv_p modifies its args.  .i attached.

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre

Attachment: product_c8.i.gz
Description: application/gunzip