From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from lists1p.gnu.org (lists1p.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-ECDSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 77126606A31C for ; Wed, 11 Nov 2026 19:37:07 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.microsoft.com header.i=@linux.microsoft.com header.a=rsa-sha256 header.s=default header.b=lY7WaBev Received: from localhost ([::1] helo=lists1p.gnu.org) by lists1p.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1xRE7Q-0002hw-Ue; Wed, 11 Nov 2026 14:36:28 -0500 Received: from eggs.gnu.org ([2001:470:142:3::10]) by lists1p.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1xRE7P-0002hV-5b for qemu-devel@nongnu.org; Wed, 11 Nov 2026 14:36:27 -0500 Received: from linux.microsoft.com ([13.77.154.182]) by eggs.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1xRE7N-0006yK-3t for qemu-devel@nongnu.org; Wed, 11 Nov 2026 14:36:26 -0500 Received: from example.com (unknown [167.220.208.43]) by linux.microsoft.com (Postfix) with ESMTPSA id 6ECBE20B7166; Wed, 11 Nov 2026 11:36:18 -0800 (PST) DKIM-Filter: OpenDKIM Filter v2.11.0 linux.microsoft.com 6ECBE20B7166 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.microsoft.com; s=default; t=1794425780; bh=PFIMXLbh6/bwWJFKvdCFfStNm4X3xyuPJxjWfUyMphk=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=lY7WaBev4eyRCK9rNr1RdawzXQ0yCqZeuR+lvDVFIR6z7Tq/PzebU5zNjryZznFgk J54y1sIaWETnxnXvco7WwRwPqt3LcNC4dR94sybY2Au6PygjZQDQAALRlSFrKB+VfE +bFtxlkPKdRwNEp2iziDFfubVqR91CyhwAveZLzI= Date: Wed, 11 Nov 2026 20:36:18 +0100 From: Magnus Kulke To: Doru =?iso-8859-1?Q?Bl=E2nzeanu?= Cc: qemu-devel@nongnu.org, "Michael S. Tsirkin" , Paolo Bonzini , Wei Liu , Wei Liu , Feng Liang , Richard Henderson , Doru =?iso-8859-1?Q?Bl=E2nzeanu?= Subject: Re: [PATCH] accel/mshv: freeze time during TSC MSR write Message-ID: References: <20261106143928.443144-1-magnuskulke@linux.microsoft.com> X-Archived-By: MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: Received-SPF: pass client-ip=13.77.154.182; envelope-from=magnuskulke@linux.microsoft.com; helo=linux.microsoft.com X-Spam_score_int: -26 X-Spam_score: -2.7 X-Spam_bar: -- X-Spam_report: (-2.7 / 5.0 requ) BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, RCVD_IN_DNSWL_LOW=-0.7, SPF_HELO_PASS=-0.001, SPF_PASS=-0.001 autolearn=ham autolearn_force=no X-Spam_action: no action X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: qemu development List-Archive: List-Post: List-Help: List-Subscribe: , Sender: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org X-Spam-Status: No, score=-15.8 required=3.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_CR_TRAILER,INCLUDES_PATCH,MAILING_LIST_MULTI,SPF_HELO_PASS, SPF_PASS autolearn=ham autolearn_force=no version=4.0.1 X-Spam-Checker-Version: SpamAssassin 4.0.1 (2024-03-26) on aws-us-west-2-korg-listserv-alma10-1 On Mon, Nov 09, 2026 at 06:22:16PM +0200, Doru Blânzeanu wrote: > On Fri, Nov 06, 2026 at 03:39:28PM +0100, Magnus Kulke wrote: > > A regression was introduced when the HV_X64_REGISTER_TSC MSR state is > > being written to the hypervisor as part of the migration framework. If > > the MSR is written without the time being frozen, the tsc_page (a > > hyper-v enlightenment) is zeroed out and cannot be used by a guest, > > which results in a performance regression, because the guest falls back > > to costly reads of HV_X64_MSR_TIME_REF_COUNT. > > > > An implementation of pre_resume_vm hook was added to handle the clock > > lifecycle, relying on the dirty_vcpu state to avoid redundant state > > writes. > > > > Fixes: d540dcc5e28c (target/i386/mshv: migrate MSRs) > > > > Signed-off-by: Magnus Kulke > > --- > > accel/mshv/mshv-all.c | 33 ++++++++++++- > > hw/i386/mshv/clock.c | 111 +++++++++++++++++++++++++++++------------- > > include/system/mshv.h | 4 +- > > 3 files changed, 113 insertions(+), 35 deletions(-) > > > > diff --git a/accel/mshv/mshv-all.c b/accel/mshv/mshv-all.c > > index d820a67b95..ae7577d407 100644 > > --- a/accel/mshv/mshv-all.c > > +++ b/accel/mshv/mshv-all.c > > @@ -34,6 +34,7 @@ > > #include "accel/accel-cpu-ops.h" > > #include "exec/cpu-common.h" > > #include "system/cpus.h" > > +#include "system/hw_accel.h" > > #include "system/runstate.h" > > #include "system/accel-blocker.h" > > #include "system/address-spaces.h" > > @@ -578,7 +579,10 @@ static int mshv_init(AccelState *as, MachineState *ms) > > > > register_savevm_live("mshv", 0, 1, &savevm_mshv, s); > > > > - mshv_clock_init(); > > + ret = mshv_clock_init(vm_fd); > > + if (ret < 0) { > > + return -1; > > + } > > > > return 0; > > } > > @@ -927,6 +931,32 @@ static void mshv_cpu_synchronize_pre_loadvm(CPUState *cpu) > > run_on_cpu(cpu, do_mshv_cpu_synchronize_pre_loadvm, RUN_ON_CPU_NULL); > > } > > > > +static void mshv_pre_resume_vm(AccelState *as, bool step_pending) > > +{ > > + MshvState *s = MSHV_STATE(as); > > + CPUState *cpu; > > + int ret, vm_fd = s->vm; > > + > > + ret = mshv_clock_prepare_resume(vm_fd); > > + if (ret < 0) { > > + error_report("Failed to prepare clock for resume"); > > + abort(); > > + } > > + > > + CPU_FOREACH(cpu) { > > + if (!cpu->vcpu_dirty) { > > + continue; > > + } > > + cpu_synchronize_post_init(cpu); > > + } > > + > > + ret = mshv_clock_finish_resume(vm_fd); > > + if (ret < 0) { > > + error_report("Failed to finish clock for resume"); > > + abort(); > > + } > > +} > > + > > static void do_mshv_cpu_synchronize(CPUState *cpu, run_on_cpu_data arg) > > { > > if (!cpu->vcpu_dirty) { > > @@ -962,6 +992,7 @@ static void mshv_accel_class_init(ObjectClass *oc, const void *data) > > ac->name = "MSHV"; > > ac->init_machine = mshv_init; > > ac->allowed = &mshv_allowed; > > + ac->pre_resume_vm = mshv_pre_resume_vm; > > } > > > > static void mshv_accel_instance_init(Object *obj) > > diff --git a/hw/i386/mshv/clock.c b/hw/i386/mshv/clock.c > > index 02c586503a..2a03d3cef9 100644 > > --- a/hw/i386/mshv/clock.c > > +++ b/hw/i386/mshv/clock.c > > @@ -29,6 +29,7 @@ > > typedef struct MshvClockState { > > uint64_t ref_time; > > bool ref_time_pending; > > + bool time_frozen; > > } MshvClockState; > > > > static MshvClockState mshv_clock; > > @@ -110,52 +111,85 @@ static int mshv_set_time_freeze(int vm_fd, int freeze) > > return 0; > > } > > > > -static void mshv_clock_vm_state_change(void *opaque, bool running, > > - RunState state) > > -{ > > - MshvClockState *s = opaque; > > - int vm_fd = mshv_state->vm; > > +static int ensure_time_frozen(int vm_fd) { > > int ret; > > > > - if (running) { > > - /* Skip if we have nothing to restore, e.g. on initial boot. */ > > - if (!s->ref_time_pending) { > > - return; > > - } > > + /* Ensure time is frozen */ > > + if (mshv_clock.time_frozen) { > > + return 0; > > + } > > > > - ret = mshv_set_time_freeze(vm_fd, 1); > > - if (ret < 0) { > > - error_report("Failed to freeze partition time on resume"); > > - abort(); > > - } > > + ret = mshv_set_time_freeze(vm_fd, 1); > > + if (ret < 0) { > > + error_report("Failed to freeze partition"); > > + return -1; > > + } > > + mshv_clock.time_frozen = true; > > > > + return 0; > > +} > > + > > +int mshv_clock_prepare_resume(int vm_fd) { > > + return ensure_time_frozen(vm_fd); > > +} > > + > > +int mshv_clock_finish_resume(int vm_fd) { > > + int ret; > > + MshvClockState *s = &mshv_clock; > > + > > + assert(mshv_clock.time_frozen); > > + > > + /* Skip if we have nothing to restore, e.g. on initial boot. */ > > + if (s->ref_time_pending) { > > /* INVARIANT: reference time can only be written if time is frozen. */ > > ret = mshv_set_reference_time(vm_fd, s->ref_time); > > if (ret < 0) { > > error_report("Failed to restore reference time on resume"); > > - abort(); > > + return -1; > > } > > + s->ref_time_pending = false; > > + } > > > > - if (mshv_set_time_freeze(vm_fd, 0) < 0) { > > - error_report("Failed to unfreeze partition time on resume"); > > - abort(); > > - } > > + ret = mshv_set_time_freeze(vm_fd, 0); > > + if (ret < 0) { > > + error_report("Failed to thaw partition"); > > + return -1; > > + } > > + mshv_clock.time_frozen = false; > > > > - s->ref_time_pending = false; > > - } else { > > - /* Skip if we already have a to-be-set ref time */ > > - if (s->ref_time_pending) { > > - return; > > - } > > + return 0; > > +} > > > > - ret = mshv_get_reference_time(vm_fd, &s->ref_time); > > - if (ret < 0) { > > - error_report("Failed to capture reference time on stop"); > > - abort(); > > - } > > +static void mshv_clock_vm_state_change(void *opaque, bool running, > > + RunState state) > > +{ > > + MshvClockState *s = opaque; > > + int vm_fd = mshv_state->vm; > > + int ret; > > > > - s->ref_time_pending = true; > > + if (running) { > > + return; > > } > > + > > + /* Ensure the time if frozen before we capture the reference time. */ > > + ret = ensure_time_frozen(vm_fd); > > + if (ret < 0) { > > + error_report("Failed to freeze partition time on stop"); > > + abort(); > > + } > > + > > + /* Skip if we already have a to-be-set ref time */ > > + if (s->ref_time_pending) { > > + return; > > + } > > + > > + ret = mshv_get_reference_time(vm_fd, &s->ref_time); > > + if (ret < 0) { > > + error_report("Failed to capture reference time on stop"); > > + abort(); > > + } > > + > > + s->ref_time_pending = true; > > } > > > > /* > > @@ -182,8 +216,19 @@ static const VMStateDescription vmstate_mshv_clock = { > > }, > > }; > > > > -void mshv_clock_init(void) > > +int mshv_clock_init(int vm_fd) > > { > > + int ret; > > + > > + ret = mshv_set_time_freeze(vm_fd, 1); > > + if (ret < 0) { > > + error_report("Failed to freeze partition time on init"); > > + return -1; > > + } > > + mshv_clock.time_frozen = true; > > + > > vmstate_register(NULL, 0, &vmstate_mshv_clock, &mshv_clock); > > qemu_add_vm_change_state_handler(mshv_clock_vm_state_change, &mshv_clock); > > + > > + return 0; > > } > > diff --git a/include/system/mshv.h b/include/system/mshv.h > > index f98ddff9b5..864f2aa34d 100644 > > --- a/include/system/mshv.h > > +++ b/include/system/mshv.h > > @@ -58,7 +58,9 @@ DECLARE_INSTANCE_CHECKER(MshvState, MSHV_STATE, > > extern MshvState *mshv_state; > > > > /* clock (partition reference time) */ > > -void mshv_clock_init(void); > > +int mshv_clock_init(int vm_fd); > > +int mshv_clock_prepare_resume(int vm_fd); > > +int mshv_clock_finish_resume(int vm_fd); > > > > /* interrupt */ > > int mshv_request_interrupt(MshvState *mshv_state, uint32_t interrupt_type, uint32_t vector, > > -- > > 2.34.1 > > Reviewed-by: Doru Blânzeanu queued in pull-20261111