From: Andrew Jones <ajones@ventanamicro.com>
To: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Cc: qemu-devel@nongnu.org, qemu-riscv@nongnu.org,
alistair.francis@wdc.com, bmeng@tinylab.org,
liweiwei@iscas.ac.cn, zhiwei_liu@linux.alibaba.com,
palmer@rivosinc.com
Subject: Re: [PATCH 01/20] target/riscv: introduce TCG AccelCPUClass
Date: Thu, 31 Aug 2023 12:17:37 +0200 [thread overview]
Message-ID: <20230831-8468b3a793485bb070b5eb6b@orel> (raw)
In-Reply-To: <20230825130853.511782-2-dbarboza@ventanamicro.com>
On Fri, Aug 25, 2023 at 10:08:34AM -0300, Daniel Henrique Barboza wrote:
> target/riscv/cpu.c needs to handle all possible accelerators (TCG and
> KVM at this moment) during both init() and realize() time. This forces
> us to resort to a lot of "if tcg" and "if kvm" throughout the code,
> which isn't wrong, but can get cluttered over time. Splitting
> acceleration specific code from cpu.c to its own file will help to
> declutter the existing code and it will also make it easier to support
> KVM/TCG only builds in the future.
>
> We'll start by adding a new subdir called 'tcg' and a new file called
> 'tcg-cpu.c'. This file will be used to introduce a new accelerator class
> for TCG acceleration in RISC-V, allowing us to center all TCG exclusive
> code in its file instead of using 'cpu.c' for everything. This design is
> inpired by the work Claudio Fontana did in x86 a few years ago in commit
> f5cc5a5c1 ("i386: split cpu accelerators from cpu.c, using
> AccelCPUClass").
>
> To avoid moving too much code at once we'll start by adding the new file
> and TCG AccelCPUClass declaration. The 'class_init' from the accel class
> will init 'tcg_ops', relieving the common riscv_cpu_class_init() from
> doing it.
>
> 'riscv_tcg_ops' is being exported from 'cpu.c' for now to avoid having
> to deal with moving code and files around right now. We'll focus on
> decoupling the realize() logic first.
>
> Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
> ---
> target/riscv/cpu.c | 5 +---
> target/riscv/cpu.h | 4 +++
> target/riscv/meson.build | 2 ++
> target/riscv/tcg/meson.build | 2 ++
> target/riscv/tcg/tcg-cpu.c | 57 ++++++++++++++++++++++++++++++++++++
> 5 files changed, 66 insertions(+), 4 deletions(-)
> create mode 100644 target/riscv/tcg/meson.build
> create mode 100644 target/riscv/tcg/tcg-cpu.c
>
> diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
> index 21ebdbf084..38dcbc4dd2 100644
> --- a/target/riscv/cpu.c
> +++ b/target/riscv/cpu.c
> @@ -2275,9 +2275,7 @@ static const struct SysemuCPUOps riscv_sysemu_ops = {
> };
> #endif
>
> -#include "hw/core/tcg-cpu-ops.h"
> -
> -static const struct TCGCPUOps riscv_tcg_ops = {
> +const struct TCGCPUOps riscv_tcg_ops = {
> .initialize = riscv_translate_init,
> .synchronize_from_tb = riscv_cpu_synchronize_from_tb,
> .restore_state_to_opc = riscv_restore_state_to_opc,
> @@ -2436,7 +2434,6 @@ static void riscv_cpu_class_init(ObjectClass *c, void *data)
> #endif
> cc->gdb_arch_name = riscv_gdb_arch_name;
> cc->gdb_get_dynamic_xml = riscv_gdb_get_dynamic_xml;
> - cc->tcg_ops = &riscv_tcg_ops;
>
> object_class_property_add(c, "mvendorid", "uint32", cpu_get_mvendorid,
> cpu_set_mvendorid, NULL, NULL);
> diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
> index 577abcd724..b84b62f84e 100644
> --- a/target/riscv/cpu.h
> +++ b/target/riscv/cpu.h
> @@ -707,6 +707,10 @@ enum riscv_pmu_event_idx {
> RISCV_PMU_EVENT_CACHE_ITLB_PREFETCH_MISS = 0x10021,
> };
>
> +/* Export tcg_ops until we move everything to tcg/tcg-cpu.c */
> +#include "hw/core/tcg-cpu-ops.h"
> +extern const struct TCGCPUOps riscv_tcg_ops;
> +
> /* CSR function table */
> extern riscv_csr_operations csr_ops[CSR_TABLE_SIZE];
>
> diff --git a/target/riscv/meson.build b/target/riscv/meson.build
> index 660078bda1..f0486183fa 100644
> --- a/target/riscv/meson.build
> +++ b/target/riscv/meson.build
> @@ -38,5 +38,7 @@ riscv_system_ss.add(files(
> 'riscv-qmp-cmds.c',
> ))
>
> +subdir('tcg')
> +
> target_arch += {'riscv': riscv_ss}
> target_softmmu_arch += {'riscv': riscv_system_ss}
> diff --git a/target/riscv/tcg/meson.build b/target/riscv/tcg/meson.build
> new file mode 100644
> index 0000000000..061df3d74a
> --- /dev/null
> +++ b/target/riscv/tcg/meson.build
> @@ -0,0 +1,2 @@
> +riscv_ss.add(when: 'CONFIG_TCG', if_true: files(
> + 'tcg-cpu.c'))
> diff --git a/target/riscv/tcg/tcg-cpu.c b/target/riscv/tcg/tcg-cpu.c
> new file mode 100644
> index 0000000000..1ad27a26aa
> --- /dev/null
> +++ b/target/riscv/tcg/tcg-cpu.c
> @@ -0,0 +1,57 @@
> +/*
> + * riscv TCG cpu class initialization
> + *
> + * Copyright (c) 2023 Ventana Micro Systems Inc.
> + *
> + * This library is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2 of the License, or (at your option) any later version.
> + *
> + * This library 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
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with this library; if not, see <http://www.gnu.org/licenses/>.
> + */
> +
> +#include "qemu/osdep.h"
> +#include "cpu.h"
> +#include "qemu/accel.h"
> +#include "hw/core/accel-cpu.h"
> +
> +static void tcg_cpu_init_ops(AccelCPUClass *accel_cpu, CPUClass *cc)
> +{
> + /*
> + * All cpus use the same set of operations.
> + * riscv_tcg_ops is being imported from cpu.c for now.
> + */
> + cc->tcg_ops = &riscv_tcg_ops;
> +}
> +
> +static void tcg_cpu_class_init(CPUClass *cc)
> +{
> + cc->init_accel_cpu = tcg_cpu_init_ops;
> +}
> +
> +static void tcg_cpu_accel_class_init(ObjectClass *oc, void *data)
> +{
> + AccelCPUClass *acc = ACCEL_CPU_CLASS(oc);
> +
> + acc->cpu_class_init = tcg_cpu_class_init;
> +}
> +
> +static const TypeInfo tcg_cpu_accel_type_info = {
> + .name = ACCEL_CPU_NAME("tcg"),
> +
> + .parent = TYPE_ACCEL_CPU,
> + .class_init = tcg_cpu_accel_class_init,
> + .abstract = true,
> +};
need blank line here
> +static void tcg_cpu_accel_register_types(void)
> +{
> + type_register_static(&tcg_cpu_accel_type_info);
> +}
> +type_init(tcg_cpu_accel_register_types);
> --
> 2.41.0
>
>
Appears to be consistent with target/i386/tcg/tcg-cpu.c, so
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Thanks,
drew
next prev parent reply other threads:[~2023-08-31 10:18 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-08-25 13:08 [PATCH 00/20] riscv: split TCG/KVM accelerators from cpu.c Daniel Henrique Barboza
2023-08-25 13:08 ` [PATCH 01/20] target/riscv: introduce TCG AccelCPUClass Daniel Henrique Barboza
2023-08-31 10:17 ` Andrew Jones [this message]
2023-08-25 13:08 ` [PATCH 02/20] target/riscv: move riscv_cpu_realize_tcg() to TCG::cpu_realizefn() Daniel Henrique Barboza
2023-08-31 10:21 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 03/20] target/riscv: move riscv_cpu_validate_set_extensions() to tcg-cpu.c Daniel Henrique Barboza
2023-08-31 10:31 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 04/20] target/riscv: move riscv_tcg_ops " Daniel Henrique Barboza
2023-08-28 16:30 ` Philippe Mathieu-Daudé
2023-08-31 10:38 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 05/20] target/riscv/cpu.c: add 'user_extension_properties' class prop Daniel Henrique Barboza
2023-08-25 13:08 ` [PATCH 06/20] target/riscv: add 'max_features' CPU flag Daniel Henrique Barboza
2023-08-25 13:08 ` [PATCH 07/20] target/riscv/cpu.c: add .instance_post_init() Daniel Henrique Barboza
2023-08-31 11:00 ` Andrew Jones
2023-09-01 20:08 ` Daniel Henrique Barboza
2023-08-25 13:08 ` [PATCH 08/20] target/riscv: move 'host' CPU declaration to kvm.c Daniel Henrique Barboza
2023-08-28 16:35 ` Philippe Mathieu-Daudé
2023-08-31 11:04 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 09/20] target/riscv/cpu.c: mark extensions arrays as 'const' Daniel Henrique Barboza
2023-08-31 11:10 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 10/20] target/riscv: move riscv_cpu_add_kvm_properties() to kvm.c Daniel Henrique Barboza
2023-08-31 11:22 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 11/20] target/riscv: introduce KVM AccelCPUClass Daniel Henrique Barboza
2023-08-28 16:38 ` Philippe Mathieu-Daudé
2023-08-29 13:16 ` Daniel Henrique Barboza
2023-08-31 11:26 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 12/20] target/riscv: move KVM only files to kvm subdir Daniel Henrique Barboza
2023-08-28 16:47 ` Philippe Mathieu-Daudé
2023-08-30 18:21 ` Daniel Henrique Barboza
2023-08-30 20:54 ` Philippe Mathieu-Daudé
2023-08-31 11:30 ` Andrew Jones
2023-09-01 17:19 ` Daniel Henrique Barboza
2023-08-25 13:08 ` [PATCH 13/20] target/riscv/kvm: refactor kvm_riscv_init_user_properties() Daniel Henrique Barboza
2023-08-31 11:34 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 14/20] target/riscv/kvm: do not use riscv_cpu_add_misa_properties() Daniel Henrique Barboza
2023-08-31 11:50 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 15/20] target/riscv/tcg: introduce tcg_cpu_instance_init() Daniel Henrique Barboza
2023-08-31 11:56 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 16/20] target/riscv/tcg: move riscv_cpu_add_misa_properties() to tcg-cpu.c Daniel Henrique Barboza
2023-08-31 12:01 ` Andrew Jones
2023-09-04 14:21 ` Daniel Henrique Barboza
2023-08-25 13:08 ` [PATCH 17/20] target/riscv/cpu.c: export isa_edata_arr[] Daniel Henrique Barboza
2023-08-31 12:06 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 18/20] target/riscv/cpu: move priv spec functions to tcg-cpu.c Daniel Henrique Barboza
2023-08-31 12:07 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 19/20] target/riscv: add 'tcg_supported' class property Daniel Henrique Barboza
2023-08-31 12:25 ` Andrew Jones
2023-08-25 13:08 ` [PATCH 20/20] target/riscv: add 'kvm_supported' " Daniel Henrique Barboza
2023-08-31 12:47 ` Andrew Jones
2023-09-01 20:57 ` Daniel Henrique Barboza
2023-09-04 9:05 ` Andrew Jones
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=20230831-8468b3a793485bb070b5eb6b@orel \
--to=ajones@ventanamicro.com \
--cc=alistair.francis@wdc.com \
--cc=bmeng@tinylab.org \
--cc=dbarboza@ventanamicro.com \
--cc=liweiwei@iscas.ac.cn \
--cc=palmer@rivosinc.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-riscv@nongnu.org \
--cc=zhiwei_liu@linux.alibaba.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).