From: Zhao Liu <zhao1.liu@intel.com>
To: "Philippe Mathieu-Daudé" <philmd@linaro.org>
Cc: qemu-devel@nongnu.org, "Thomas Huth" <thuth@redhat.com>,
"Richard Henderson" <richard.henderson@linaro.org>,
"Paolo Bonzini" <pbonzini@redhat.com>,
"Alistair Francis" <alistair.francis@wdc.com>,
qemu-arm@nongnu.org, qemu-riscv@nongnu.org,
"Edgar E. Iglesias" <edgar.iglesias@gmail.com>,
qemu-ppc@nongnu.org, "Eduardo Habkost" <eduardo@habkost.net>,
"Michael S. Tsirkin" <mst@redhat.com>,
qemu-s390x@nongnu.org, "Peter Maydell" <peter.maydell@linaro.org>,
"Alex Bennée" <alex.bennee@linaro.org>,
"Zhao Liu" <zhao1.liu@intel.com>
Subject: Re: [PATCH 00/19] cpus: Step toward removing global 'first_cpu'
Date: Mon, 23 Oct 2023 21:59:40 +0800 [thread overview]
Message-ID: <ZTZ8TEGmc4WXGRnp@intel.com> (raw)
In-Reply-To: <20231020163643.86105-1-philmd@linaro.org>
Hi Philippe,
On Fri, Oct 20, 2023 at 06:36:22PM +0200, Philippe Mathieu-Daudé wrote:
> Date: Fri, 20 Oct 2023 18:36:22 +0200
> From: Philippe Mathieu-Daudé <philmd@linaro.org>
> Subject: [PATCH 00/19] cpus: Step toward removing global 'first_cpu'
> X-Mailer: git-send-email 2.41.0
>
> Heterogeneous machines have different type of CPU.
> In that context, both 'first_cpu' and qemu_get_cpu(index)
> don't make much sense.
Could you talk more about what the issue that heterogeneous machines
will meet on these 2 things?
This question relates to when one should need to pay attention to CPU
type.
>
> This series first add a 'filter CPU type' argument to
> qemu_get_cpu(), so the behavior respects what we currently
> expect, then replace 'first_cpu' by a qemu_get_cpu(0) call.
Considering that the current uses of first_cpu and qemu_get_cpu(index)
are CPU type-agnostic, would it be more straightforward to convert most
of the current uses to qemu_get_cpu(index, NULL)?
Regards,
Zhao
>
> Testing pass, but I'm still unsure about performance impact.
>
> With this series applied, what is left is:
>
> [OK]
>
> - hw/core/generic-loader.c:127: s->cpu = qemu_get_cpu(s->cpu_num, NULL);
> - linux-user/syscall.c:8990: if (CPU_NEXT(first_cpu)) {
> - monitor/hmp-cmds-target.c:106: cs = vcpu >= 0 ? qemu_get_cpu(vcpu, NULL) : mon_get_cpu(mon);
> - stats/stats-hmp-cmds.c:150: CPUState *cpu = qemu_get_cpu(cpu_index, NULL);
> - system/cpus.c:754: cpu = qemu_get_cpu(cpu_index, NULL);
>
> [likely OK, we don't support heterogeneous accelerators]
>
> - accel/kvm/kvm-all.c:3581: kvm_arch_on_sigbus_vcpu(first_cpu, code, addr);
> - accel/kvm/kvm-all.c:4255: if (first_cpu) {
> - accel/kvm/kvm-all.c:4258: query_stats_schema_vcpu(first_cpu, &stats_args);
>
> [need some toughts...]
>
> - accel/tcg/tcg-accel-ops-rr.c:114: qemu_cond_wait_iothread(first_cpu->halt_cond);
> - accel/tcg/tcg-accel-ops-rr.c:200: while (first_cpu->stopped) {
> - accel/tcg/tcg-accel-ops-rr.c:201: qemu_cond_wait_iothread(first_cpu->halt_cond);
> - accel/tcg/tcg-accel-ops-rr.c:212: cpu = first_cpu;
> - accel/tcg/tcg-accel-ops-rr.c:242: cpu = first_cpu;
> - accel/tcg/tcg-accel-ops-rr.c:336: cpu->thread_id = first_cpu->thread_id;
> - accel/tcg/tcg-accel-ops-rr.c:92: if (!rr_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
> - dump/win_dump.c:115: if (cpu_memory_rw_debug(first_cpu,
> - dump/win_dump.c:128: if (cpu_read_ptr(x64, first_cpu,
> - dump/win_dump.c:135: if (cpu_memory_rw_debug(first_cpu, KiBugcheckData,
> - dump/win_dump.c:210: if (cpu_memory_rw_debug(first_cpu,
> - dump/win_dump.c:258: if (cpu_read_ptr(x64, first_cpu,
> - dump/win_dump.c:265: if (cpu_memory_rw_debug(first_cpu,
> - dump/win_dump.c:286: if (cpu_read_ptr(x64, first_cpu,
> - dump/win_dump.c:294: if (cpu_read_ptr(x64, first_cpu,
> - dump/win_dump.c:378: if (cpu_memory_rw_debug(first_cpu, Context,
> - dump/win_dump.c:384: if (cpu_memory_rw_debug(first_cpu, Context,
> - dump/win_dump.c:400: if (cpu_memory_rw_debug(first_cpu, saved_ctx[i].addr,
> - dump/win_dump.c:410: X86CPU *first_x86_cpu = X86_CPU(first_cpu);
> - gdbstub/gdbstub.c:1521: cc = CPU_GET_CLASS(first_cpu);
> - gdbstub/gdbstub.c:298: CPUState *cpu = first_cpu;
> - gdbstub/system.c:338: if (!first_cpu) {
> - hw/acpi/cpu.c:138: if (!cdev->cpu || cdev->cpu == first_cpu) {
> - hw/acpi/cpu.c:150: if (!cdev->cpu || cdev->cpu == first_cpu) {
> - hw/acpi/cpu.c:671: if (CPU(arch_ids->cpus[i].cpu) != first_cpu) {
> - hw/core/generic-loader.c:134: s->cpu = first_cpu;
> - monitor/hmp-cmds-target.c:39: cpu = qemu_get_cpu(cpu_index, NULL);
> - monitor/hmp-cmds-target.c:62: if (!first_cpu) {
> - monitor/hmp-cmds-target.c:65: monitor_set_cpu(mon, first_cpu->cpu_index);
> - monitor/hmp-cmds-target.c:66: cpu = first_cpu;
> - replay/replay-events.c:130: qemu_cpu_kick(first_cpu);
> - system/cpu-timers.c:254: } else if (first_cpu) {
> - system/cpu-timers.c:263: async_run_on_cpu(first_cpu, do_nothing, RUN_ON_CPU_NULL);
> - system/memory_mapping.c:315: first_paging_enabled_cpu = find_paging_enabled_cpu(first_cpu);
> - system/qtest.c:548: address_space_write(first_cpu->as, ...
> - system/qtest.c:553: address_space_write(first_cpu->as, ...
> - system/qtest.c:558: address_space_write(first_cpu->as, ...
> - system/qtest.c:563: address_space_write(first_cpu->as, ...
> - system/qtest.c:582: address_space_read(first_cpu->as, ...
> - system/qtest.c:587: address_space_read(first_cpu->as, ...
> - system/qtest.c:592: address_space_read(first_cpu->as, ...
> - system/qtest.c:596: address_space_read(first_cpu->as, ...
> - system/qtest.c:617: address_space_read(first_cpu->as, ...
> - system/qtest.c:643: address_space_read(first_cpu->as, ...
> - system/qtest.c:678: address_space_write(first_cpu->as, ...
> - system/qtest.c:701: address_space_write(first_cpu->as, ...
> - system/qtest.c:735: address_space_write(first_cpu->as, ...
> - tests/qtest/fuzz/generic_fuzz.c:241: mr1 = address_space_translate(first_cpu->as,
> - tests/qtest/fuzz/qtest_wrappers.c:110: address_space_read(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:122: address_space_read(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:134: address_space_read(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:146: address_space_read(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:157: address_space_write(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:167: address_space_write(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:177: address_space_write(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:187: address_space_write(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:197: address_space_read(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:207: address_space_read(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:218: address_space_write(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:229: address_space_write(first_cpu->as, ...
> - tests/qtest/fuzz/qtest_wrappers.c:242: address_space_write(first_cpu->as, ...
>
> Regards,
>
> Phil.
>
> Based-on: <20231013140116.255-1-philmd@linaro.org>
> "target: Make 'cpu-qom.h' really target agnostic" v2
>
> Philippe Mathieu-Daudé (19):
> cpus: Add argument to qemu_get_cpu() to filter CPUs by QOM type
> cpus: Filter for target specific CPU (generic)
> cpus: Filter for target specific CPU (arm)
> cpus: Filter for target specific CPU (loongarch)
> cpus: Filter for target specific CPU (mips)
> cpus: Filter for target specific CPU (s390x)
> cpus: Filter for target specific CPU (riscv)
> cpus: Filter for target specific CPU (ppc)
> cpus: Filter for target specific CPU (x86)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_ARM_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_POWERPC_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_MIPS_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_M68K_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_S390X_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_RISCV_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_TRICORE_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_SUPERH_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_RX_CPU)
> cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_X86_CPU)
>
> include/hw/core/cpu.h | 3 ++-
> target/mips/internal.h | 2 +-
> cpu-common.c | 5 ++++-
> hw/arm/aspeed.c | 5 +++--
> hw/arm/bananapi_m2u.c | 3 ++-
> hw/arm/boot.c | 14 +++++++-------
> hw/arm/exynos4_boards.c | 6 ++++--
> hw/arm/fsl-imx7.c | 2 +-
> hw/arm/highbank.c | 3 ++-
> hw/arm/microbit.c | 3 ++-
> hw/arm/mps2-tz.c | 3 ++-
> hw/arm/mps2.c | 3 ++-
> hw/arm/msf2-som.c | 3 ++-
> hw/arm/musca.c | 3 ++-
> hw/arm/netduino2.c | 3 ++-
> hw/arm/netduinoplus2.c | 2 +-
> hw/arm/olimex-stm32-h405.c | 2 +-
> hw/arm/orangepi.c | 3 ++-
> hw/arm/pxa2xx_gpio.c | 2 +-
> hw/arm/realview.c | 5 +++--
> hw/arm/sbsa-ref.c | 7 ++++---
> hw/arm/stellaris.c | 3 ++-
> hw/arm/stm32vldiscovery.c | 2 +-
> hw/arm/vexpress.c | 5 +++--
> hw/arm/virt-acpi-build.c | 2 +-
> hw/arm/virt.c | 25 ++++++++++++++-----------
> hw/arm/xilinx_zynq.c | 3 ++-
> hw/arm/xlnx-versal-virt.c | 2 +-
> hw/core/generic-loader.c | 2 +-
> hw/cpu/a15mpcore.c | 4 ++--
> hw/cpu/a9mpcore.c | 2 +-
> hw/hyperv/hyperv.c | 2 +-
> hw/i386/kvm/clock.c | 4 ++--
> hw/i386/kvm/xen_evtchn.c | 8 ++++----
> hw/i386/kvmvapic.c | 3 ++-
> hw/i386/microvm.c | 2 +-
> hw/i386/pc.c | 7 ++++---
> hw/i386/pc_piix.c | 3 ++-
> hw/i386/x86.c | 2 +-
> hw/intc/arm_gicv3_common.c | 3 ++-
> hw/intc/arm_gicv3_cpuif.c | 2 +-
> hw/intc/arm_gicv3_kvm.c | 2 +-
> hw/intc/m68k_irqc.c | 2 +-
> hw/intc/mips_gic.c | 2 +-
> hw/intc/riscv_aclint.c | 2 +-
> hw/intc/sh_intc.c | 6 ++++--
> hw/intc/sifive_plic.c | 4 ++--
> hw/isa/lpc_ich9.c | 2 +-
> hw/loongarch/virt.c | 10 +++++-----
> hw/m68k/mcf5206.c | 2 +-
> hw/mips/bootloader.c | 3 ++-
> hw/mips/cps.c | 5 +++--
> hw/mips/loongson3_bootp.c | 2 +-
> hw/mips/loongson3_virt.c | 7 ++++---
> hw/mips/malta.c | 2 +-
> hw/misc/mips_cpc.c | 4 ++--
> hw/ppc/e500.c | 4 ++--
> hw/ppc/ppc.c | 4 ++--
> hw/ppc/ppce500_spin.c | 2 +-
> hw/ppc/prep_systemio.c | 2 +-
> hw/ppc/spapr.c | 11 ++++++-----
> hw/ppc/spapr_caps.c | 10 +++++-----
> hw/ppc/spapr_rtas.c | 2 +-
> hw/ppc/spapr_vof.c | 2 +-
> hw/ppc/vof.c | 3 ++-
> hw/riscv/boot.c | 4 ++--
> hw/riscv/opentitan.c | 4 ++--
> hw/rx/rx-gdbsim.c | 4 ++--
> hw/s390x/ipl.c | 4 ++--
> hw/s390x/s390-virtio-ccw.c | 2 +-
> hw/tricore/triboard.c | 2 +-
> monitor/hmp-cmds-target.c | 4 ++--
> stats/stats-hmp-cmds.c | 2 +-
> system/cpus.c | 2 +-
> target/arm/arch_dump.c | 6 +++---
> target/i386/arch_dump.c | 6 +++---
> target/i386/kvm/kvm.c | 6 +++---
> target/i386/kvm/xen-emu.c | 15 ++++++++-------
> target/i386/monitor.c | 2 +-
> target/i386/tcg/sysemu/fpu_helper.c | 4 ++--
> target/mips/cpu.c | 2 +-
> target/mips/tcg/sysemu/cp0_helper.c | 10 +++++-----
> target/mips/tcg/sysemu/tlb_helper.c | 2 +-
> target/ppc/arch_dump.c | 4 ++--
> target/riscv/arch_dump.c | 4 ++--
> target/s390x/cpu_models.c | 10 +++++-----
> 86 files changed, 199 insertions(+), 163 deletions(-)
>
> --
> 2.41.0
>
prev parent reply other threads:[~2023-10-23 13:49 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-10-20 16:36 [PATCH 00/19] cpus: Step toward removing global 'first_cpu' Philippe Mathieu-Daudé
2023-10-20 16:36 ` [RFC PATCH 01/19] cpus: Add argument to qemu_get_cpu() to filter CPUs by QOM type Philippe Mathieu-Daudé
2023-10-20 17:14 ` Peter Maydell
2023-10-20 17:29 ` Philippe Mathieu-Daudé
2023-10-20 17:40 ` Peter Maydell
2023-10-23 14:21 ` Zhao Liu
2023-10-23 15:12 ` David Woodhouse
2023-10-20 16:36 ` [PATCH 02/19] cpus: Filter for target specific CPU (generic) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 03/19] cpus: Filter for target specific CPU (arm) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 04/19] cpus: Filter for target specific CPU (loongarch) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 05/19] cpus: Filter for target specific CPU (mips) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 06/19] cpus: Filter for target specific CPU (s390x) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 07/19] cpus: Filter for target specific CPU (riscv) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 08/19] cpus: Filter for target specific CPU (ppc) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 09/19] cpus: Filter for target specific CPU (x86) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 10/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_ARM_CPU) Philippe Mathieu-Daudé
2023-10-20 16:53 ` Cédric Le Goater
2023-10-20 16:36 ` [PATCH 11/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_POWERPC_CPU) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 12/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_MIPS_CPU) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 13/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_M68K_CPU) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 14/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_S390X_CPU) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 15/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_RISCV_CPU) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 16/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_TRICORE_CPU) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 17/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_SUPERH_CPU) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 18/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_RX_CPU) Philippe Mathieu-Daudé
2023-10-20 16:36 ` [PATCH 19/19] cpus: Replace first_cpu by qemu_get_cpu(0, TYPE_X86_CPU) Philippe Mathieu-Daudé
2023-10-23 13:59 ` Zhao Liu [this message]
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=ZTZ8TEGmc4WXGRnp@intel.com \
--to=zhao1.liu@intel.com \
--cc=alex.bennee@linaro.org \
--cc=alistair.francis@wdc.com \
--cc=edgar.iglesias@gmail.com \
--cc=eduardo@habkost.net \
--cc=mst@redhat.com \
--cc=pbonzini@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=philmd@linaro.org \
--cc=qemu-arm@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@nongnu.org \
--cc=qemu-riscv@nongnu.org \
--cc=qemu-s390x@nongnu.org \
--cc=richard.henderson@linaro.org \
--cc=thuth@redhat.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.