qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Paolo Bonzini <pbonzini@redhat.com>
To: qemu-devel@nongnu.org
Cc: Richard Henderson <richard.henderson@linaro.org>,
	Peter Xu <peterx@redhat.com>, Igor Mammedov <imammedo@redhat.com>
Subject: [PULL 09/61] accel: use store_release/load_acquire for cross-thread exit_request
Date: Sat, 13 Sep 2025 10:08:50 +0200	[thread overview]
Message-ID: <20250913080943.11710-10-pbonzini@redhat.com> (raw)
In-Reply-To: <20250913080943.11710-1-pbonzini@redhat.com>

Reads and writes cpu->exit_request do not use a load-acquire/store-release
pair right now, but this means that cpu_exit() may not write cpu->exit_request
after any flags that are read by the vCPU thread.

Probably everything is protected one way or the other by the BQL, because
cpu->exit_request leads to the slow path, where the CPU thread often takes
the BQL (for example, to go to sleep by waiting on the BQL-protected
cpu->halt_cond); but it's not clear, so use load-acquire/store-release
consistently.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c          | 19 +++++++++----------
 accel/tcg/cpu-exec.c         |  7 +++++--
 accel/tcg/tcg-accel-ops-rr.c | 11 +++++++++--
 hw/core/cpu-common.c         |  3 ++-
 target/i386/nvmm/nvmm-all.c  |  5 ++---
 target/i386/whpx/whpx-all.c  |  3 ++-
 6 files changed, 29 insertions(+), 19 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index f36dfe33492..bd9e5e3886d 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -3029,10 +3029,6 @@ static void kvm_eat_signals(CPUState *cpu)
 
     if (kvm_immediate_exit) {
         qatomic_set(&cpu->kvm_run->immediate_exit, 0);
-        /* Write kvm_run->immediate_exit before the cpu->exit_request
-         * write in kvm_cpu_exec.
-         */
-        smp_wmb();
         return;
     }
 
@@ -3187,7 +3183,8 @@ int kvm_cpu_exec(CPUState *cpu)
         }
 
         kvm_arch_pre_run(cpu, run);
-        if (qatomic_read(&cpu->exit_request)) {
+        /* Corresponding store-release is in cpu_exit. */
+        if (qatomic_load_acquire(&cpu->exit_request)) {
             trace_kvm_interrupt_exit_request();
             /*
              * KVM requires us to reenter the kernel after IO exits to complete
@@ -3197,13 +3194,15 @@ int kvm_cpu_exec(CPUState *cpu)
             kvm_cpu_kick_self();
         }
 
-        /* Read cpu->exit_request before KVM_RUN reads run->immediate_exit.
-         * Matching barrier in kvm_eat_signals.
-         */
-        smp_rmb();
-
         run_ret = kvm_vcpu_ioctl(cpu, KVM_RUN, 0);
 
+        /*
+         * After writing cpu->exit_request, cpu_exit() sends a signal that writes
+         * kvm->run->immediate_exit.  The signal is already happening after the
+         * write to cpu->exit_request so, if KVM read kvm->run->immediate_exit
+         * as true, cpu->exit_request will always read as true.
+         */
+
         attrs = kvm_arch_post_run(cpu, run);
 
 #ifdef KVM_HAVE_MCE_INJECTION
diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index 508d2d2d9e2..f838535d111 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -851,8 +851,11 @@ static inline bool cpu_handle_interrupt(CPUState *cpu,
     }
 #endif /* !CONFIG_USER_ONLY */
 
-    /* Finally, check if we need to exit to the main loop.  */
-    if (unlikely(qatomic_read(&cpu->exit_request)) || icount_exit_request(cpu)) {
+    /*
+     * Finally, check if we need to exit to the main loop.
+     * The corresponding store-release is in cpu_exit.
+     */
+    if (unlikely(qatomic_load_acquire(&cpu->exit_request)) || icount_exit_request(cpu)) {
         qatomic_set(&cpu->exit_request, 0);
         if (cpu->exception_index == -1) {
             cpu->exception_index = EXCP_INTERRUPT;
diff --git a/accel/tcg/tcg-accel-ops-rr.c b/accel/tcg/tcg-accel-ops-rr.c
index 6eec5c9eee9..e8b0e370a8d 100644
--- a/accel/tcg/tcg-accel-ops-rr.c
+++ b/accel/tcg/tcg-accel-ops-rr.c
@@ -242,10 +242,17 @@ static void *rr_cpu_thread_fn(void *arg)
             cpu = first_cpu;
         }
 
-        while (cpu && cpu_work_list_empty(cpu) && !cpu->exit_request) {
-            /* Store rr_current_cpu before evaluating cpu_can_run().  */
+        while (cpu && cpu_work_list_empty(cpu)) {
+            /*
+             * Store rr_current_cpu before evaluating cpu->exit_request.
+             * Pairs with rr_kick_next_cpu().
+             */
             qatomic_set_mb(&rr_current_cpu, cpu);
 
+            /* Pairs with store-release in cpu_exit.  */
+            if (qatomic_load_acquire(&cpu->exit_request)) {
+                break;
+            }
             current_cpu = cpu;
 
             qemu_clock_enable(QEMU_CLOCK_VIRTUAL,
diff --git a/hw/core/cpu-common.c b/hw/core/cpu-common.c
index 152abc9024b..42463e6258d 100644
--- a/hw/core/cpu-common.c
+++ b/hw/core/cpu-common.c
@@ -74,7 +74,8 @@ void cpu_reset_interrupt(CPUState *cpu, int mask)
 
 void cpu_exit(CPUState *cpu)
 {
-    qatomic_set(&cpu->exit_request, 1);
+    /* Ensure cpu_exec will see the reason why the exit request was set.  */
+    qatomic_store_release(&cpu->exit_request, true);
     /* Ensure cpu_exec will see the exit request after TCG has exited.  */
     smp_wmb();
     qatomic_set(&cpu->neg.icount_decr.u16.high, -1);
diff --git a/target/i386/nvmm/nvmm-all.c b/target/i386/nvmm/nvmm-all.c
index e1151b04c6e..10bd51d9b59 100644
--- a/target/i386/nvmm/nvmm-all.c
+++ b/target/i386/nvmm/nvmm-all.c
@@ -743,7 +743,8 @@ nvmm_vcpu_loop(CPUState *cpu)
 
         nvmm_vcpu_pre_run(cpu);
 
-        if (qatomic_read(&cpu->exit_request)) {
+        /* Corresponding store-release is in cpu_exit. */
+        if (qatomic_load_acquire(&cpu->exit_request)) {
 #if NVMM_USER_VERSION >= 2
             nvmm_vcpu_stop(vcpu);
 #else
@@ -751,8 +752,6 @@ nvmm_vcpu_loop(CPUState *cpu)
 #endif
         }
 
-        /* Read exit_request before the kernel reads the immediate exit flag */
-        smp_rmb();
         ret = nvmm_vcpu_run(mach, vcpu);
         if (ret == -1) {
             error_report("NVMM: Failed to exec a virtual processor,"
diff --git a/target/i386/whpx/whpx-all.c b/target/i386/whpx/whpx-all.c
index c09a0a64f22..2106c29c3a0 100644
--- a/target/i386/whpx/whpx-all.c
+++ b/target/i386/whpx/whpx-all.c
@@ -1714,7 +1714,8 @@ static int whpx_vcpu_run(CPUState *cpu)
         if (exclusive_step_mode == WHPX_STEP_NONE) {
             whpx_vcpu_pre_run(cpu);
 
-            if (qatomic_read(&cpu->exit_request)) {
+            /* Corresponding store-release is in cpu_exit. */
+            if (qatomic_load_acquire(&cpu->exit_request)) {
                 whpx_vcpu_kick(cpu);
             }
         }
-- 
2.51.0



  parent reply	other threads:[~2025-09-13  8:11 UTC|newest]

Thread overview: 73+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-09-13  8:08 [PULL 00/61] CPU, Rust, x86 changes for 2025-09-13 Paolo Bonzini
2025-09-13  8:08 ` [PULL 01/61] target/ppc: limit cpu_interrupt_exittb to system emulation Paolo Bonzini
2025-09-13  8:08 ` [PULL 02/61] target/sparc: limit cpu_check_irqs " Paolo Bonzini
2025-09-13  8:08 ` [PULL 03/61] target/i386: limit a20 " Paolo Bonzini
2025-09-13  8:08 ` [PULL 04/61] target-arm: remove uses of cpu_interrupt() for user-mode emulation Paolo Bonzini
2025-09-13  8:08 ` [PULL 05/61] user-exec: remove cpu_interrupt() stub Paolo Bonzini
2025-09-13  8:08 ` [PULL 06/61] treewide: clear bits of cs->interrupt_request with cpu_reset_interrupt() Paolo Bonzini
2025-09-13  8:08 ` [PULL 07/61] cpu-common: use atomic access for interrupt_request Paolo Bonzini
2025-09-13  8:08 ` [PULL 08/61] cpus: document that qemu_cpu_kick() can be used for BQL-less operation Paolo Bonzini
2025-09-13  8:08 ` Paolo Bonzini [this message]
2025-09-13  8:08 ` [PULL 10/61] accel: use atomic accesses for exit_request Paolo Bonzini
2025-09-13  8:08 ` [PULL 11/61] accel/tcg: create a thread-kick function for TCG Paolo Bonzini
2025-09-13  8:08 ` [PULL 12/61] accel/tcg: inline cpu_exit() Paolo Bonzini
2025-09-13  8:08 ` [PULL 13/61] cpus: remove TCG-ism from cpu_exit() Paolo Bonzini
2025-09-13  8:08 ` [PULL 14/61] cpus: properly kick CPUs out of inner execution loop Paolo Bonzini
2025-09-13  8:08 ` [PULL 15/61] treewide: rename qemu_wait_io_event/qemu_wait_io_event_common Paolo Bonzini
2025-09-13  8:08 ` [PULL 16/61] bsd-user, linux-user: introduce qemu_process_cpu_events Paolo Bonzini
2025-09-13  8:08 ` [PULL 17/61] cpus: clear exit_request in qemu_process_cpu_events Paolo Bonzini
2025-09-13  8:08 ` [PULL 18/61] accel: make all calls to qemu_process_cpu_events look the same Paolo Bonzini
2025-09-13  8:09 ` [PULL 19/61] tcg/user: do not set exit_request gratuitously Paolo Bonzini
2025-09-13  8:09 ` [PULL 20/61] ci: temporarily remove rust from Ubuntu Paolo Bonzini
2025-09-13  8:09 ` [PULL 21/61] configure: bump Meson to 1.9.0 for use with Rust Paolo Bonzini
2025-09-22 14:07   ` Peter Maydell
2025-09-22 15:14     ` Paolo Bonzini
2025-09-13  8:09 ` [PULL 22/61] meson, cargo: require Rust 1.83.0 Paolo Bonzini
2025-09-13  8:09 ` [PULL 23/61] rust: add missing const markers for MSRV==1.83.0 Paolo Bonzini
2025-09-13  8:09 ` [PULL 24/61] rust: use inline const expressions Paolo Bonzini
2025-09-13  8:09 ` [PULL 25/61] rust: add qdev Device derive macro Paolo Bonzini
2025-09-13  8:09 ` [PULL 26/61] rust: vmstate: convert to use builder pattern Paolo Bonzini
2025-09-13  8:09 ` [PULL 27/61] rust: vmstate: use const_refs_to_static Paolo Bonzini
2025-09-13  8:09 ` [PULL 28/61] rust: qdev: const_refs_to_static Paolo Bonzini
2025-09-13  8:09 ` [PULL 29/61] docs/rust: update msrv Paolo Bonzini
2025-09-13  8:09 ` [PULL 30/61] rust: remove unused global qemu "allocator" Paolo Bonzini
2025-09-13  8:09 ` [PULL 31/61] rust: add workspace authors Paolo Bonzini
2025-09-13  8:09 ` [PULL 32/61] rust: move vmstate_clock!() to qdev module Paolo Bonzini
2025-09-13  8:09 ` [PULL 33/61] rust: move VMState handling to QOM module Paolo Bonzini
2025-09-13  8:09 ` [PULL 34/61] rust: move Cell vmstate impl Paolo Bonzini
2025-09-13  8:09 ` [PULL 35/61] rust: split Rust-only "common" crate Paolo Bonzini
2025-09-13  8:09 ` [PULL 36/61] rust: make build.rs generic over various ./rust/projects Paolo Bonzini
2025-09-13  8:09 ` [PULL 37/61] rust: split "util" crate Paolo Bonzini
2025-09-13  8:09 ` [PULL 38/61] rust: split "migration" crate Paolo Bonzini
2025-09-13  8:09 ` [PULL 39/61] rust: split "bql" crate Paolo Bonzini
2025-09-13  8:09 ` [PULL 40/61] rust: split "qom" crate Paolo Bonzini
2025-09-13  8:09 ` [PULL 41/61] rust: split "chardev" crate Paolo Bonzini
2025-09-13  8:09 ` [PULL 42/61] rust: split "system" crate Paolo Bonzini
2025-09-13  8:09 ` [PULL 43/61] rust: split "hwcore" crate Paolo Bonzini
2025-09-13  8:09 ` [PULL 44/61] rust: rename qemu_api_macros -> qemu_macros Paolo Bonzini
2025-09-13  8:09 ` [PULL 45/61] rust/hpet: drop now unneeded qemu_api dep Paolo Bonzini
2025-09-13  8:09 ` [PULL 46/61] rust/pl011: drop dependency on qemu_api Paolo Bonzini
2025-09-13  8:09 ` [PULL 47/61] rust: repurpose qemu_api -> tests Paolo Bonzini
2025-09-13  8:09 ` [PULL 48/61] rust: re-export qemu_macros internal helper in "bits" Paolo Bonzini
2025-09-13  8:09 ` [PULL 49/61] rust: re-export qemu macros from common/qom/hwcore Paolo Bonzini
2025-09-13  8:09 ` [PULL 50/61] docs: update rust.rst Paolo Bonzini
2025-09-13  8:09 ` [PULL 51/61] rust: meson: remove unnecessary complication in device crates Paolo Bonzini
2025-09-13  8:09 ` [PULL 52/61] rust: do not inline do_init_io Paolo Bonzini
2025-09-13  8:09 ` [PULL 53/61] hpet: guard IRQ handling with BQL Paolo Bonzini
2025-09-13  8:09 ` [PULL 54/61] i386/cpu: Enable SMM cpu address space under KVM Paolo Bonzini
2025-09-18 16:24   ` Michael Tokarev
2025-09-22 15:16     ` Paolo Bonzini
2025-09-26 17:48   ` Peter Maydell
2025-09-28  6:51     ` Xiaoyao Li
2025-09-13  8:09 ` [PULL 55/61] target/i386: Define enum X86ASIdx for x86's address spaces Paolo Bonzini
2025-09-13  8:09 ` [PULL 56/61] multiboot: Fix the split lock Paolo Bonzini
2025-09-13  8:09 ` [PULL 57/61] i386/kvm: Get X86MachineState in kvm_arch_init() without the cast check Paolo Bonzini
2025-09-13  8:09 ` [PULL 58/61] i386/kvm: Drop KVM_CAP_X86_SMM check in kvm_arch_init() Paolo Bonzini
2025-09-13  8:09 ` [PULL 59/61] accel/kvm: Switch to check KVM_CAP_GUEST_MEMFD and KVM_CAP_USER_MEMORY2 on VM Paolo Bonzini
2025-09-13  8:09 ` [PULL 60/61] accel/kvm: Zero out mem explicitly in kvm_set_user_memory_region() Paolo Bonzini
2025-09-13  8:09 ` [PULL 61/61] accel/kvm: Set guest_memfd_offset to non-zero value only when guest_memfd is valid Paolo Bonzini
2025-09-13  9:37 ` [PULL 00/61] CPU, Rust, x86 changes for 2025-09-13 Peter Maydell
2025-09-16 14:37   ` Peter Maydell
2025-09-16 14:53     ` Paolo Bonzini
2025-09-17 16:33 ` Richard Henderson
2025-09-18 12:38   ` Xiaoyao Li

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=20250913080943.11710-10-pbonzini@redhat.com \
    --to=pbonzini@redhat.com \
    --cc=imammedo@redhat.com \
    --cc=peterx@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=richard.henderson@linaro.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).