From: Paolo Bonzini <pbonzini@redhat.com>
To: qemu-devel@nongnu.org
Subject: [PATCH v3 00/19] accel, cpus: clean up cpu->exit_request
Date: Mon, 8 Sep 2025 12:12:22 +0200 [thread overview]
Message-ID: <20250908101241.2110068-1-pbonzini@redhat.com> (raw)
cpu->exit_request is used to kick vCPU threads into qemu_wait_io_event().
The code that handles the signaling of cpu->exit_request is messy, mostly
due to no one ever taking a look at it as a whole. In fact already in commit
4b8523ee896 ("kvm: First step to push iothread lock out of inner run loop",
2015-07-01), the read of cpu->exit_request was placed outside the BQL
critical section without much attention to ordering; and it only got
worse from that point, in no small part due to a young and naive me.
This series is complementary to the cpu->interrupt_request cleanups
that Igor did in "memory: reintroduce BQL-free fine-grained PIO/MMIO"
(https://lore.kernel.org/qemu-devel/20250808120137.2208800-1-imammedo@redhat.com/T/),
and in fact includes some extra changes to interrupt_request at the
beginning.
Paolo
Paolo Bonzini (19):
target/ppc: limit cpu_interrupt_exittb to system emulation
target/sparc: limit cpu_check_irqs to system emulation
target/i386: limit a20 to system emulation
target-arm: remove uses of cpu_interrupt() for user-mode emulation
user-exec: remove cpu_interrupt() stub
treewide: clear bits of cs->interrupt_request with cpu_reset_interrupt()
cpu-common: use atomic access for interrupt_request
cpus: document that qemu_cpu_kick() can be used for BQL-less operation
accel: use store_release/load_acquire for cross-thread exit_request
accel: use atomic accesses for exit_request
accel/tcg: create a thread-kick function for TCG
accel/tcg: inline cpu_exit()
cpus: remove TCG-ism from cpu_exit()
cpus: properly kick CPUs out of inner execution loop
treewide: rename qemu_wait_io_event/qemu_wait_io_event_common
bsd-user, linux-user: introduce qemu_process_cpu_events
cpus: clear exit_request in qemu_process_cpu_events
accel: make all calls to qemu_process_cpu_events look the same
tcg/user: do not set exit_request gratuitously
v2->v3:
- [accel: use store_release/load_acquire for cross-thread exit_request]
improve comments for memory barriers, match rr_kick_next_cpu() more
accurately
- [cpus: properly kick CPUs out of inner execution loop]
fix comment
- [treewide: rename qemu_wait_io_event/qemu_wait_io_event_common]
new patch; new names used throughout in patches 16-19
docs/devel/tcg-icount.rst | 2 +-
accel/tcg/tcg-accel-ops-mttcg.h | 3 -
accel/tcg/tcg-accel-ops.h | 1 +
bsd-user/aarch64/target_arch_cpu.h | 2 +-
bsd-user/arm/target_arch_cpu.h | 2 +-
bsd-user/i386/target_arch_cpu.h | 2 +-
bsd-user/riscv/target_arch_cpu.h | 2 +-
bsd-user/x86_64/target_arch_cpu.h | 2 +-
include/hw/core/cpu.h | 22 +-
include/system/cpus.h | 3 +-
target/arm/internals.h | 5 +
accel/dummy-cpus.c | 2 +-
accel/hvf/hvf-accel-ops.c | 2 +-
accel/kvm/kvm-accel-ops.c | 3 +-
accel/kvm/kvm-all.c | 23 +-
accel/tcg/cpu-exec.c | 34 ++-
accel/tcg/tcg-accel-ops-mttcg.c | 12 +-
accel/tcg/tcg-accel-ops-rr.c | 55 ++--
accel/tcg/tcg-accel-ops.c | 4 +-
accel/tcg/user-exec.c | 11 +-
bsd-user/main.c | 5 -
cpu-common.c | 3 +-
hw/core/cpu-common.c | 19 +-
hw/core/cpu-system.c | 2 +-
hw/ppc/ppc.c | 2 +
hw/ppc/spapr_hcall.c | 7 +-
hw/ppc/spapr_rtas.c | 2 +-
linux-user/aarch64/cpu_loop.c | 2 +-
linux-user/alpha/cpu_loop.c | 2 +-
linux-user/arm/cpu_loop.c | 2 +-
linux-user/hexagon/cpu_loop.c | 2 +-
linux-user/hppa/cpu_loop.c | 2 +-
linux-user/i386/cpu_loop.c | 2 +-
linux-user/loongarch64/cpu_loop.c | 2 +-
linux-user/m68k/cpu_loop.c | 2 +-
linux-user/main.c | 5 -
linux-user/microblaze/cpu_loop.c | 2 +-
linux-user/mips/cpu_loop.c | 2 +-
linux-user/openrisc/cpu_loop.c | 2 +-
linux-user/ppc/cpu_loop.c | 2 +-
linux-user/riscv/cpu_loop.c | 2 +-
linux-user/s390x/cpu_loop.c | 2 +-
linux-user/sh4/cpu_loop.c | 2 +-
linux-user/sparc/cpu_loop.c | 2 +-
linux-user/xtensa/cpu_loop.c | 2 +-
replay/replay-events.c | 3 +-
system/cpu-timers.c | 6 +-
system/cpus.c | 15 +-
target/arm/cpu-irq.c | 381 ++++++++++++++++++++++++++++
target/arm/cpu.c | 370 ---------------------------
target/arm/el2-stubs.c | 37 +++
target/arm/helper.c | 4 +
target/arm/tcg/mte_helper.c | 2 +-
target/avr/helper.c | 4 +-
target/i386/helper.c | 2 +
target/i386/hvf/x86hvf.c | 8 +-
target/i386/kvm/hyperv.c | 1 -
target/i386/kvm/kvm.c | 20 +-
target/i386/nvmm/nvmm-accel-ops.c | 8 +-
target/i386/nvmm/nvmm-all.c | 19 +-
target/i386/tcg/system/seg_helper.c | 13 +-
target/i386/tcg/system/svm_helper.c | 2 +-
target/i386/whpx/whpx-accel-ops.c | 6 +-
target/i386/whpx/whpx-all.c | 23 +-
target/openrisc/sys_helper.c | 2 +-
target/ppc/helper_regs.c | 2 +
target/rx/helper.c | 4 +-
target/s390x/tcg/excp_helper.c | 2 +-
target/sparc/int32_helper.c | 2 +
target/sparc/int64_helper.c | 2 +
target/arm/meson.build | 2 +
71 files changed, 645 insertions(+), 566 deletions(-)
create mode 100644 target/arm/cpu-irq.c
create mode 100644 target/arm/el2-stubs.c
--
2.51.0
next reply other threads:[~2025-09-08 10:15 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-08 10:12 Paolo Bonzini [this message]
2025-09-08 10:12 ` [PATCH 01/19] target/ppc: limit cpu_interrupt_exittb to system emulation Paolo Bonzini
2025-09-08 10:12 ` [PATCH 02/19] target/sparc: limit cpu_check_irqs " Paolo Bonzini
2025-09-08 10:12 ` [PATCH 03/19] target/i386: limit a20 " Paolo Bonzini
2025-09-08 10:12 ` [PATCH 04/19] target-arm: remove uses of cpu_interrupt() for user-mode emulation Paolo Bonzini
2025-09-08 10:12 ` [PATCH 05/19] user-exec: remove cpu_interrupt() stub Paolo Bonzini
2025-09-08 10:12 ` [PATCH 06/19] treewide: clear bits of cs->interrupt_request with cpu_reset_interrupt() Paolo Bonzini
2025-09-08 10:12 ` [PATCH 07/19] cpu-common: use atomic access for interrupt_request Paolo Bonzini
2025-09-08 10:12 ` [PATCH 08/19] cpus: document that qemu_cpu_kick() can be used for BQL-less operation Paolo Bonzini
2025-09-08 10:12 ` [PATCH 09/19] accel: use store_release/load_acquire for cross-thread exit_request Paolo Bonzini
2025-09-08 10:12 ` [PATCH 10/19] accel: use atomic accesses for exit_request Paolo Bonzini
2025-09-08 10:12 ` [PATCH 11/19] accel/tcg: create a thread-kick function for TCG Paolo Bonzini
2025-09-08 10:12 ` [PATCH 12/19] accel/tcg: inline cpu_exit() Paolo Bonzini
2025-09-08 10:12 ` [PATCH 13/19] cpus: remove TCG-ism from cpu_exit() Paolo Bonzini
2025-09-08 10:12 ` [PATCH 14/19] cpus: properly kick CPUs out of inner execution loop Paolo Bonzini
2025-09-10 7:13 ` Richard Henderson
2025-09-08 10:12 ` [PATCH 15/19] treewide: rename qemu_wait_io_event/qemu_wait_io_event_common Paolo Bonzini
2025-09-10 7:14 ` Richard Henderson
2025-09-08 10:12 ` [PATCH 16/19] bsd-user, linux-user: introduce qemu_process_cpu_events Paolo Bonzini
2025-09-08 10:12 ` [PATCH 17/19] cpus: clear exit_request in qemu_process_cpu_events Paolo Bonzini
2025-09-08 10:12 ` [PATCH 18/19] accel: make all calls to qemu_process_cpu_events look the same Paolo Bonzini
2025-09-08 10:12 ` [PATCH 19/19] tcg/user: do not set exit_request gratuitously Paolo Bonzini
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=20250908101241.2110068-1-pbonzini@redhat.com \
--to=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
/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).