From: Richard Henderson <richard.henderson@linaro.org>
To: Taylor Simpson <tsimpson@quicinc.com>, qemu-devel@nongnu.org
Cc: ale@rev.ng, riku.voipio@iki.fi, philmd@redhat.com,
laurent@vivier.eu, aleksandar.m.mail@gmail.com
Subject: Re: [RFC PATCH v3 07/34] Hexagon (target/hexagon) scalar core helpers
Date: Wed, 26 Aug 2020 07:16:32 -0700 [thread overview]
Message-ID: <7014720f-eda6-50f9-4594-8510d5745d63@linaro.org> (raw)
In-Reply-To: <1597765847-16637-8-git-send-email-tsimpson@quicinc.com>
On 8/18/20 8:50 AM, Taylor Simpson wrote:
> The majority of helpers are generated. Define the helper functions needed
> then include the generated file
>
> Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
> ---
> target/hexagon/helper.h | 33 ++++
> target/hexagon/op_helper.c | 368 +++++++++++++++++++++++++++++++++++++++++++++
> 2 files changed, 401 insertions(+)
> create mode 100644 target/hexagon/helper.h
> create mode 100644 target/hexagon/op_helper.c
>
> diff --git a/target/hexagon/helper.h b/target/hexagon/helper.h
> new file mode 100644
> index 0000000..48b1917
> --- /dev/null
> +++ b/target/hexagon/helper.h
> @@ -0,0 +1,33 @@
> +/*
> + * Copyright(c) 2019-2020 Qualcomm Innovation Center, Inc. All Rights Reserved.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, see <http://www.gnu.org/licenses/>.
> + */
> +
> +DEF_HELPER_2(raise_exception, noreturn, env, i32)
> +DEF_HELPER_1(debug_start_packet, void, env)
> +DEF_HELPER_3(debug_check_store_width, void, env, int, int)
> +DEF_HELPER_3(debug_commit_end, void, env, int, int)
> +DEF_HELPER_3(merge_inflight_store1s, s32, env, s32, s32)
> +DEF_HELPER_3(merge_inflight_store1u, s32, env, s32, s32)
> +DEF_HELPER_3(merge_inflight_store2s, s32, env, s32, s32)
> +DEF_HELPER_3(merge_inflight_store2u, s32, env, s32, s32)
> +DEF_HELPER_3(merge_inflight_store4s, s32, env, s32, s32)
> +DEF_HELPER_3(merge_inflight_store4u, s32, env, s32, s32)
> +DEF_HELPER_3(merge_inflight_store8u, s64, env, s32, s64)
Please use DEF_HELPER_FLAGS_N when possible.
You should be able to know when
(1) No exceptions are possible, and nothing is touched except the inputs. In
this case use TCG_CALL_NO_RWG_SE.
(2) No exceptions are possible, and nothing in env is touched for which you
have created a tcg variable in hexagon_translate_init(). In this case use
TCG_CALL_NO_RWG.
(3) Exceptions are possible, and no tcg variables are touched on the
non-exceptional path. In this case use TCG_CALL_NO_WG.
> +static inline void log_pred_write(CPUHexagonState *env, int pnum,
> + target_ulong val)
You might be better off letting the compiler decide about inlining.
> +/*
> + * Handle mem_noshuf
> + *
> + * This occurs when there is a load that might need data forwarded
> + * from an inflight store in slot 1. Note that the load and store
> + * might have different sizes, so we can't simply compare the
> + * addresses. We merge only the bytes that overlap (if any).
> + */
> +static int64_t merge_bytes(CPUHexagonState *env, target_ulong load_addr,
> + int64_t load_data, uint32_t load_width)
> +{
> + /* Don't do anything if slot 1 was cancelled */
> + const int store_slot = 1;
> + if (env->slot_cancelled & (1 << store_slot)) {
> + return load_data;
> + }
> +
> + int store_width = env->mem_log_stores[store_slot].width;
> + target_ulong store_addr = env->mem_log_stores[store_slot].va;
> + union {
> + uint8_t bytes[8];
> + uint32_t data32;
> + uint64_t data64;
> + } retdata, storedata;
Red flag here. This is assuming that the host and the target are both
little-endian.
> + int bigmask = ((-load_width) & (-store_width));
> + if ((load_addr & bigmask) != (store_addr & bigmask)) {
Huh? This doesn't detect overlap. Counter example:
load_addr = 0, load_width = 4,
store_addr = 1, store_width = 1.
bigmask = -4 & -1 -> -4.
(0 & -4) != (1 & -4) -> 0 != 1
> + while ((i < load_width) && (j < store_width)) {
> + retdata.bytes[i] = storedata.bytes[j];
> + i++;
> + j++;
> + }
> + return retdata.data64;
This most definitely requires host-endian adjustment.
> +/* Helpful for printing intermediate values within instructions */
> +void HELPER(debug_value)(CPUHexagonState *env, int32_t value)
> +{
> + HEX_DEBUG_LOG("value = 0x%x\n", value);
> +}
> +
> +void HELPER(debug_value_i64)(CPUHexagonState *env, int64_t value)
> +{
> + HEX_DEBUG_LOG("value_i64 = 0x%lx\n", value);
> +}
I think we need to lose all of this.
There are other ways to debug TCG.
r~
next prev parent reply other threads:[~2020-08-26 14:17 UTC|newest]
Thread overview: 122+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-08-18 15:50 [RFC PATCH v3 00/34] Hexagon patch series Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 01/34] Hexagon Update MAINTAINERS file Taylor Simpson
2020-08-26 1:55 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 02/34] Hexagon (target/hexagon) README Taylor Simpson
2020-08-26 2:06 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 03/34] Hexagon (include/elf.h) ELF machine definition Taylor Simpson
2020-08-26 2:06 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 04/34] Hexagon (target/hexagon) scalar core definition Taylor Simpson
2020-08-26 13:35 ` Richard Henderson
2020-08-26 23:51 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 05/34] Hexagon (target/hexagon) register names Taylor Simpson
2020-08-26 13:39 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 06/34] Hexagon (disas) disassembler Taylor Simpson
2020-08-26 13:52 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 07/34] Hexagon (target/hexagon) scalar core helpers Taylor Simpson
2020-08-26 14:16 ` Richard Henderson [this message]
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 08/34] Hexagon (target/hexagon) GDB Stub Taylor Simpson
2020-08-26 14:17 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 09/34] Hexagon (target/hexagon) architecture types Taylor Simpson
2020-08-26 14:19 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 10/34] Hexagon (target/hexagon) instruction and packet types Taylor Simpson
2020-08-26 14:22 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 11/34] Hexagon (target/hexagon) register fields Taylor Simpson
2020-08-26 14:29 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 12/34] Hexagon (target/hexagon) instruction attributes Taylor Simpson
2020-08-26 14:34 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 13/34] Hexagon (target/hexagon) register map Taylor Simpson
2020-08-26 14:36 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 14/34] Hexagon (target/hexagon) instruction/packet decode Taylor Simpson
2020-08-26 15:06 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 15/34] Hexagon (target/hexagon) instruction printing Taylor Simpson
2020-08-26 15:08 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 16/34] Hexagon (target/hexagon) utility functions Taylor Simpson
2020-08-26 15:10 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 17/34] Hexagon (target/hexagon/imported) arch import - macro definitions Taylor Simpson
2020-08-26 15:17 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 18/34] Hexagon (target/hexagon/imported) arch import - instruction semantics Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 19/34] Hexagon (target/hexagon/imported) arch import - instruction encoding Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 20/34] Hexagon (target/hexagon) generator phase 1 - C preprocessor for semantics Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 21/34] Hexagon (target/hexagon) generator phase 2 - generate header files Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 22/34] Hexagon (target/hexagon) generator phase 3 - C preprocessor for decode tree Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 23/34] Hexagon (target/hexagon) generater phase 4 - " Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 24/34] Hexagon (target/hexagon) opcode data structures Taylor Simpson
2020-08-26 15:25 ` Richard Henderson
2020-08-26 23:52 ` Taylor Simpson
2020-08-27 4:05 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 25/34] Hexagon (target/hexagon) macros to interface with the generator Taylor Simpson
2020-08-29 0:49 ` Richard Henderson
2020-08-30 20:30 ` Taylor Simpson
2020-08-30 20:59 ` Richard Henderson
2020-08-30 21:20 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 26/34] Hexagon (target/hexagon) macros referenced in instruction semantics Taylor Simpson
2020-08-29 1:16 ` Richard Henderson
2020-08-30 20:23 ` Taylor Simpson
2020-08-30 21:06 ` Richard Henderson
2020-10-08 15:00 ` Taylor Simpson
2020-10-08 17:30 ` Richard Henderson
2020-10-08 18:51 ` Taylor Simpson
2020-10-08 20:02 ` Richard Henderson
2020-10-08 20:54 ` Taylor Simpson
2020-10-09 12:59 ` Richard Henderson
2020-10-09 16:02 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 27/34] Hexagon (target/hexagon) instruction classes Taylor Simpson
2020-08-29 1:37 ` Richard Henderson
2020-08-30 20:04 ` Taylor Simpson
2020-08-30 20:43 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 28/34] Hexagon (target/hexagon) TCG generation helpers Taylor Simpson
2020-08-29 1:48 ` Richard Henderson
2020-08-30 19:53 ` Taylor Simpson
2020-08-30 20:52 ` Richard Henderson
2020-08-30 21:38 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 29/34] Hexagon (target/hexagon) TCG generation Taylor Simpson
2020-08-29 1:58 ` Richard Henderson
2020-08-30 19:49 ` Taylor Simpson
2020-08-18 15:50 ` [RFC PATCH v3 30/34] Hexagon (target/hexagon) TCG for instructions with multiple definitions Taylor Simpson
2020-08-29 2:02 ` Richard Henderson
2020-08-30 19:48 ` Taylor Simpson
2020-08-30 21:13 ` Richard Henderson
2020-08-30 21:30 ` Taylor Simpson
2020-08-30 23:26 ` Richard Henderson
2020-08-31 17:08 ` Taylor Simpson
2020-08-31 17:29 ` Richard Henderson
2020-08-31 18:14 ` Taylor Simpson
2020-08-31 19:20 ` Richard Henderson
2020-08-31 23:10 ` Taylor Simpson
2020-09-01 2:40 ` Richard Henderson
2020-09-01 4:17 ` Taylor Simpson
2020-09-24 2:56 ` Taylor Simpson
2020-09-24 15:03 ` Richard Henderson
2020-09-24 17:18 ` Taylor Simpson
2020-09-24 19:04 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 31/34] Hexagon (target/hexagon) translation Taylor Simpson
2020-08-29 2:49 ` Richard Henderson
2020-08-30 19:37 ` Taylor Simpson
2020-08-30 23:08 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 32/34] Hexagon (linux-user/hexagon) Linux user emulation Taylor Simpson
2020-08-29 2:59 ` Richard Henderson
2020-08-18 15:50 ` [RFC PATCH v3 33/34] Hexagon (tests/tcg/hexagon) TCG tests Taylor Simpson
2020-08-29 3:05 ` Richard Henderson
2020-09-01 9:57 ` Alessandro Di Federico
2020-08-18 15:50 ` [RFC PATCH v3 34/34] Hexagon build infrastructure Taylor Simpson
2020-08-29 3:19 ` Richard Henderson
2020-09-24 2:35 ` Taylor Simpson
2020-09-25 16:59 ` Philippe Mathieu-Daudé
2020-08-18 16:32 ` [RFC PATCH v3 00/34] Hexagon patch series no-reply
2020-08-29 3:27 ` Richard Henderson
2020-08-30 20:47 ` Taylor Simpson
2020-08-30 23:33 ` Richard Henderson
2020-08-31 17:57 ` Taylor Simpson
2020-08-31 20:43 ` Richard Henderson
2020-08-31 23:48 ` Taylor Simpson
2020-09-07 9:49 ` Rob Landley
2020-09-15 0:41 ` [EXT] " Brian Cain
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=7014720f-eda6-50f9-4594-8510d5745d63@linaro.org \
--to=richard.henderson@linaro.org \
--cc=ale@rev.ng \
--cc=aleksandar.m.mail@gmail.com \
--cc=laurent@vivier.eu \
--cc=philmd@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=riku.voipio@iki.fi \
--cc=tsimpson@quicinc.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 NNTP newsgroup(s).