From: will schmidt <will_schmidt@vnet.ibm.com>
To: Bill Schmidt <wschmidt@linux.ibm.com>, gcc-patches@gcc.gnu.org
Cc: segher@kernel.crashing.org
Subject: Re: [PATCH 48/55] rs6000: Builtin expansion, part 5
Date: Tue, 27 Jul 2021 16:07:00 -0500 [thread overview]
Message-ID: <4285eb69fc2c51e7c2235facda1b9400cb7e806c.camel@vnet.ibm.com> (raw)
In-Reply-To: <f75fbd2a4ffe51e08451743d9bb94d8bfce497e5.1623941442.git.wschmidt@linux.ibm.com>
On Thu, 2021-06-17 at 10:19 -0500, Bill Schmidt via Gcc-patches wrote:
> 2021-06-17 Bill Schmidt <wschmidt@linux.ibm.com>
>
Hi,
> gcc/
> * config/rs6000/rs6000-call.c (new_mma_expand_builtin):
> Implement.
Ok,
> ---
> gcc/config/rs6000/rs6000-call.c | 103 ++++++++++++++++++++++++++++++++
> 1 file changed, 103 insertions(+)
>
> diff --git a/gcc/config/rs6000/rs6000-call.c b/gcc/config/rs6000/rs6000-call.c
> index 981eabc1187..e1b685fb874 100644
> --- a/gcc/config/rs6000/rs6000-call.c
> +++ b/gcc/config/rs6000/rs6000-call.c
> @@ -14962,6 +14962,109 @@ static rtx
> new_mma_expand_builtin (tree exp, rtx target, insn_code icode,
> rs6000_gen_builtins fcode)
> {
> + tree fndecl = TREE_OPERAND (CALL_EXPR_FN (exp), 0);
> + tree arg;
> + call_expr_arg_iterator iter;
> + const struct insn_operand_data *insn_op;
> + rtx op[MAX_MMA_OPERANDS];
> + unsigned nopnds = 0;
> + bool void_func = TREE_TYPE (TREE_TYPE (fndecl)) == void_type_node;
> + machine_mode tmode = VOIDmode;
> +
> + if (!void_func)
> + {
> + tmode = insn_data[icode].operand[0].mode;
> + if (!target
> + || GET_MODE (target) != tmode
> + || !(*insn_data[icode].operand[0].predicate) (target, tmode))
> + target = gen_reg_rtx (tmode);
> + op[nopnds++] = target;
> + }
> + else
> + target = const0_rtx;
> +
> + FOR_EACH_CALL_EXPR_ARG (arg, iter, exp)
> + {
> + if (arg == error_mark_node)
> + return const0_rtx;
> +
> + rtx opnd;
> + insn_op = &insn_data[icode].operand[nopnds];
> + if (TREE_CODE (arg) == ADDR_EXPR
> + && MEM_P (DECL_RTL (TREE_OPERAND (arg, 0))))
> + opnd = DECL_RTL (TREE_OPERAND (arg, 0));
> + else
> + opnd = expand_normal (arg);
> +
> + if (!(*insn_op->predicate) (opnd, insn_op->mode))
> + {
> + if (!strcmp (insn_op->constraint, "n"))
> + {
> + if (!CONST_INT_P (opnd))
> + error ("argument %d must be an unsigned literal", nopnds);
> + else
> + error ("argument %d is an unsigned literal that is "
> + "out of range", nopnds);
> + return const0_rtx;
> + }
> + opnd = copy_to_mode_reg (insn_op->mode, opnd);
> + }
> +
> + /* Some MMA instructions have INOUT accumulator operands, so force
> + their target register to be the same as their input register. */
> + if (!void_func
> + && nopnds == 1
> + && !strcmp (insn_op->constraint, "0")
> + && insn_op->mode == tmode
> + && REG_P (opnd)
> + && (*insn_data[icode].operand[0].predicate) (opnd, tmode))
> + target = op[0] = opnd;
> +
> + op[nopnds++] = opnd;
> + }
> +
> + rtx pat;
> + switch (nopnds)
> + {
> + case 1:
> + pat = GEN_FCN (icode) (op[0]);
> + break;
> + case 2:
> + pat = GEN_FCN (icode) (op[0], op[1]);
> + break;
> + case 3:
> + /* The ASSEMBLE builtin source operands are reversed in little-endian
> + mode, so reorder them. */
> + if (fcode == RS6000_BIF_ASSEMBLE_PAIR_V_INTERNAL && !WORDS_BIG_ENDIAN)
> + std::swap (op[1], op[2]);
> + pat = GEN_FCN (icode) (op[0], op[1], op[2]);
> + break;
> + case 4:
> + pat = GEN_FCN (icode) (op[0], op[1], op[2], op[3]);
> + break;
> + case 5:
> + /* The ASSEMBLE builtin source operands are reversed in little-endian
> + mode, so reorder them. */
I'd be tempted to consolidate the source operand reversal comments for
RS6000_BIF_ASSEMBLE_PAIR_V_INTERNAL and
RS6000_BIF_ASSEMBLE_ACC_INTERNAL.. up at the start of the case
statement.. but actually i think this makes sense as-is.
Ok.
> + if (fcode == RS6000_BIF_ASSEMBLE_ACC_INTERNAL && !WORDS_BIG_ENDIAN)
> + {
> + std::swap (op[1], op[4]);
> + std::swap (op[2], op[3]);
> + }
> + pat = GEN_FCN (icode) (op[0], op[1], op[2], op[3], op[4]);
> + break;
> + case 6:
> + pat = GEN_FCN (icode) (op[0], op[1], op[2], op[3], op[4], op[5]);
> + break;
> + case 7:
> + pat = GEN_FCN (icode) (op[0], op[1], op[2], op[3], op[4], op[5], op[6]);
> + break;
> + default:
> + gcc_unreachable ();
> + }
> + if (!pat)
> + return NULL_RTX;
> + emit_insn (pat);
> +
Ok,
lgtm,
thanks
-Will
> return target;
> }
>
next prev parent reply other threads:[~2021-07-27 21:07 UTC|newest]
Thread overview: 110+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-17 15:18 [PATCHv3 00/55] Replace the Power target-specific builtin machinery Bill Schmidt
2021-06-17 15:18 ` [PATCH 01/55] Support scanning of build-time GC roots in gengtype Bill Schmidt
2021-06-17 15:18 ` [PATCH 02/55] rs6000: Initial create of rs6000-gen-builtins.c Bill Schmidt
2021-06-17 15:18 ` [PATCH 03/55] rs6000: Add initial input files Bill Schmidt
2021-06-17 15:18 ` [PATCH 04/55] rs6000: Add file support and functions for diagnostic support Bill Schmidt
2021-06-17 15:18 ` [PATCH 05/55] rs6000: Add helper functions for parsing Bill Schmidt
2021-07-09 19:32 ` will schmidt
2021-07-14 22:58 ` Segher Boessenkool
2021-07-14 23:32 ` Segher Boessenkool
2021-06-17 15:18 ` [PATCH 06/55] rs6000: Add functions for matching types, part 1 of 3 Bill Schmidt
2021-06-17 15:18 ` [PATCH 07/55] rs6000: Add functions for matching types, part 2 " Bill Schmidt
2021-06-17 15:18 ` [PATCH 08/55] rs6000: Add functions for matching types, part 3 " Bill Schmidt
2021-06-17 15:18 ` [PATCH 09/55] rs6000: Red-black tree implementation for balanced tree search Bill Schmidt
2021-06-17 15:18 ` [PATCH 10/55] rs6000: Main function with stubs for parsing and output Bill Schmidt
2021-07-19 19:15 ` Segher Boessenkool
2021-07-20 22:19 ` Bill Schmidt
2021-07-20 23:22 ` Segher Boessenkool
2021-07-21 1:51 ` Bill Schmidt
2021-07-21 15:43 ` Segher Boessenkool
2021-07-21 16:08 ` Bill Schmidt
2021-07-21 16:16 ` Bill Schmidt
2021-06-17 15:18 ` [PATCH 11/55] rs6000: Parsing built-in input file, part 1 of 3 Bill Schmidt
2021-07-19 20:39 ` Segher Boessenkool
2021-06-17 15:18 ` [PATCH 12/55] rs6000: Parsing built-in input file, part 2 " Bill Schmidt
2021-07-19 22:07 ` Segher Boessenkool
2021-06-17 15:18 ` [PATCH 13/55] rs6000: Parsing built-in input file, part 3 " Bill Schmidt
2021-07-19 22:13 ` Segher Boessenkool
2021-06-17 15:18 ` [PATCH 14/55] rs6000: Parsing of overload input file Bill Schmidt
2021-07-19 23:09 ` Segher Boessenkool
2021-06-17 15:18 ` [PATCH 15/55] rs6000: Build and store function type identifiers Bill Schmidt
2021-07-20 0:04 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 16/55] rs6000: Write output to the builtin definition include file Bill Schmidt
2021-07-20 23:27 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 17/55] rs6000: Write output to the builtins header file Bill Schmidt
2021-07-20 23:40 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 18/55] rs6000: Write output to the builtins init file, part 1 of 3 Bill Schmidt
2021-07-20 23:51 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 19/55] rs6000: Write output to the builtins init file, part 2 " Bill Schmidt
2021-07-20 23:53 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 20/55] rs6000: Write output to the builtins init file, part 3 " Bill Schmidt
2021-07-21 17:08 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 21/55] rs6000: Write static initializations for built-in table Bill Schmidt
2021-07-21 17:14 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 22/55] rs6000: Write static initializations for overload tables Bill Schmidt
2021-07-21 17:40 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 23/55] rs6000: Incorporate new builtins code into the build machinery Bill Schmidt
2021-07-21 18:58 ` Segher Boessenkool
2021-07-27 3:26 ` Bill Schmidt
2021-07-27 14:23 ` Segher Boessenkool
2021-07-27 17:38 ` Bill Schmidt
2021-06-17 15:19 ` [PATCH 24/55] rs6000: Add gengtype handling to " Bill Schmidt
2021-06-17 15:19 ` [PATCH 25/55] rs6000: Add the rest of the [altivec] stanza to the builtins file Bill Schmidt
2021-06-17 15:19 ` [PATCH 26/55] rs6000: Add VSX builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 27/55] rs6000: Add available-everywhere and ancient builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 28/55] rs6000: Add power7 and power7-64 builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 29/55] rs6000: Add power8-vector builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 30/55] rs6000: Add Power9 builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 31/55] rs6000: Add more type nodes to support builtin processing Bill Schmidt
2021-06-17 15:19 ` [PATCH 32/55] rs6000: Add Power10 builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 33/55] rs6000: Add MMA builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 34/55] rs6000: Add miscellaneous builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 35/55] rs6000: Add Cell builtins Bill Schmidt
2021-06-17 15:19 ` [PATCH 36/55] rs6000: Add remaining overloads Bill Schmidt
2021-06-17 15:19 ` [PATCH 37/55] rs6000: Execute the automatic built-in initialization code Bill Schmidt
2021-06-17 15:19 ` [PATCH 38/55] rs6000: Darwin builtin support Bill Schmidt
2021-06-17 15:19 ` [PATCH 39/55] rs6000: Add sanity to V2DI_type_node definitions Bill Schmidt
2021-06-17 15:19 ` [PATCH 40/55] rs6000: Always initialize vector_pair and vector_quad nodes Bill Schmidt
2021-06-17 15:19 ` [PATCH 41/55] rs6000: Handle overloads during program parsing Bill Schmidt
2021-06-17 15:19 ` [PATCH 42/55] rs6000: Handle gimple folding of target built-ins Bill Schmidt
2021-07-28 21:21 ` will schmidt
2021-07-29 12:42 ` Bill Schmidt
2021-08-02 13:31 ` Bill Schmidt
2021-08-02 23:43 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 43/55] rs6000: Support for vectorizing built-in functions Bill Schmidt
2021-06-17 15:19 ` [PATCH 44/55] rs6000: Builtin expansion, part 1 Bill Schmidt
2021-07-27 21:06 ` will schmidt
2021-07-28 3:30 ` Bill Schmidt
2021-06-17 15:19 ` [PATCH 45/55] rs6000: Builtin expansion, part 2 Bill Schmidt
2021-07-27 21:06 ` will schmidt
2021-06-17 15:19 ` [PATCH 46/55] rs6000: Builtin expansion, part 3 Bill Schmidt
2021-07-27 21:06 ` will schmidt
2021-08-03 23:40 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 47/55] rs6000: Builtin expansion, part 4 Bill Schmidt
2021-07-27 21:06 ` will schmidt
2021-08-03 23:46 ` Segher Boessenkool
2021-08-04 0:34 ` Segher Boessenkool
2021-08-12 16:17 ` Bill Schmidt
2021-06-17 15:19 ` [PATCH 48/55] rs6000: Builtin expansion, part 5 Bill Schmidt
2021-07-27 21:07 ` will schmidt [this message]
2021-06-17 15:19 ` [PATCH 49/55] rs6000: Builtin expansion, part 6 Bill Schmidt
2021-07-27 21:07 ` will schmidt
2021-07-28 20:38 ` Bill Schmidt
2021-06-17 15:19 ` [PATCH 50/55] rs6000: Update rs6000_builtin_decl Bill Schmidt
2021-07-27 21:08 ` will schmidt
2021-08-04 0:38 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 51/55] rs6000: Miscellaneous uses of rs6000_builtin_decls_x Bill Schmidt
2021-07-27 21:08 ` will schmidt
2021-06-17 15:19 ` [PATCH 52/55] rs6000: Debug support Bill Schmidt
2021-07-27 21:07 ` will schmidt
2021-08-04 0:49 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 53/55] rs6000: Update altivec.h for automated interfaces Bill Schmidt
2021-07-27 21:07 ` will schmidt
2021-07-28 20:58 ` Bill Schmidt
2021-08-04 0:58 ` Segher Boessenkool
2021-06-17 15:19 ` [PATCH 54/55] rs6000: Test case adjustments Bill Schmidt
2021-06-17 15:19 ` [PATCH 55/55] rs6000: Enable the new builtin support Bill Schmidt
2021-07-27 21:07 ` will schmidt
2021-06-25 15:25 ` [PATCHv3 00/55] Replace the Power target-specific builtin machinery Bill Schmidt
2021-07-13 13:52 ` Bill Schmidt
-- strict thread matches above, loose matches on Subject: below --
2021-06-08 18:26 [PATCHv2 " Bill Schmidt
2021-06-08 18:26 ` [PATCH 48/55] rs6000: Builtin expansion, part 5 Bill Schmidt
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4285eb69fc2c51e7c2235facda1b9400cb7e806c.camel@vnet.ibm.com \
--to=will_schmidt@vnet.ibm.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=segher@kernel.crashing.org \
--cc=wschmidt@linux.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).