Re: [stage1][PATCH] Lower VEC_COND_EXPR into internal functions.

["Martin Liška" <mliska@suse.cz>]

Adding Segher to CC, he can help us.

Martin

On 5/21/20 2:51 PM, Martin Liška wrote:
> Hi.
> 
> Back to this I noticed that ppc64le target build is broken due to:
> 
> g++  -fno-PIE -c   -g   -DIN_GCC  -DCROSS_DIRECTORY_STRUCTURE   -fno-exceptions -fno-rtti -fasynchronous-unwind-tables -W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual -Wmissing-format-attribute -Woverloaded-virtual -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -fno-common  -DHAVE_CONFIG_H -I. -I. -I/home/marxin/Programming/gcc/gcc -I/home/marxin/Programming/gcc/gcc/. -I/home/marxin/Programming/gcc/gcc/../include -I/home/marxin/Programming/gcc/gcc/../libcpp/include  -I/home/marxin/Programming/gcc/gcc/../libdecnumber -I/home/marxin/Programming/gcc/gcc/../libdecnumber/dpd -I../libdecnumber -I/home/marxin/Programming/gcc/gcc/../libbacktrace   -o insn-emit.o -MT insn-emit.o -MMD -MP -MF ./.deps/insn-emit.TPo insn-emit.c
> /home/marxin/Programming/gcc/gcc/config/rs6000/vector.md:357:11: error: vcondv4sfv4sf cannot FAIL
>    357 |     FAIL;
>        |           ^
> /home/marxin/Programming/gcc/gcc/config/rs6000/vector.md:357:11: error: vcondv2dfv2df cannot FAIL
>    357 |     FAIL;
>        |           ^
> /home/marxin/Programming/gcc/gcc/config/rs6000/vector.md:374:11: error: vcondv16qiv16qi cannot FAIL
>    374 |     FAIL;
>        |           ^
> /home/marxin/Programming/gcc/gcc/config/rs6000/vector.md:374:11: error: vcondv8hiv8hi cannot FAIL
>    374 |     FAIL;
>        |           ^
> ...
> 
> 
> which is caused by the 4 added optabs:
> 
> +DEF_INTERNAL_OPTAB_FN (VCOND, 0, vcond, vec_cond)
> +DEF_INTERNAL_OPTAB_FN (VCONDU, 0, vcondu, vec_condu)
> +DEF_INTERNAL_OPTAB_FN (VCONDEQ, 0, vcondeq, vec_condeq)
> +DEF_INTERNAL_OPTAB_FN (VCOND_MASK, 0, vcond_mask, vec_cond_mask)
> 
> looking at the generator:
> 
> Breakpoint 6, gen_expand (info=0x7fffffffe160) at /home/marxin/Programming/gcc/gcc/genemit.c:516
> 516          if (find_optab (&p, XSTR (expand, 0)))
> (gdb) bt
> #0  emit_c_code (code=0x7fa0f0 "{\n  if (rs6000_emit_vector_cond_expr (operands[0], operands[1], operands[2],\n\t\t\t\t    operands[3], operands[4], operands[5]))\n    DONE;\n  else\n    FAIL;\n}", can_fail_p=false, name=0x7fa190 "vcondv4sfv4sf")
>      at /home/marxin/Programming/gcc/gcc/genemit.c:306
> #1  0x00000000004039b5 in gen_expand (info=0x7fffffffe160) at /home/marxin/Programming/gcc/gcc/genemit.c:522
> #2  0x0000000000404912 in main (argc=4, argv=0x7fffffffe288) at /home/marxin/Programming/gcc/gcc/genemit.c:916
> 
> I get there due to:
> 
> B- │516               if (find_optab (&p, XSTR (expand, 0)))│
>     │517                 {                                   │
>     │518                   gcc_assert (p.op < NUM_OPTABS);   │
>     │519                   if (nofail_optabs[p.op])          │
>     │520                     can_fail_p = false;             │
>     │521                 }                                   │
> 
> 
> #define DEF_INTERNAL_OPTAB_FN(NAME, FLAGS, OPTAB, TYPE) \
>    nofail_optabs[OPTAB##_optab] = true;
> #include "internal-fn.def"
> 
> Any hint what's bad? Note that x86_64-linux-gnu is fine.
> Do I miss a target hook?
> 
> Martin