From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from foss.arm.com (foss.arm.com [217.140.110.172]) by smtp.subspace.kernel.org (Postfix) with ESMTP id A692B348C6A; Mon, 8 Jun 2026 11:15:08 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=217.140.110.172 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1780917310; cv=none; b=tDbtC6RgG7V5yhIURwmr2aYu5QSG4ZSO6hpa+s96KjQY0uy6cDSTgCJASZXaiKRNtZzLFGaJTj9lTJssfNpiUYmJNUU1tbfMMsCQg4kJ9a/7L17/YMeb4fNpyz6YKGfr31dCpqKBmKV1Rv5DBbPnJOprXFgyzWHH8DS6cKcEDc4= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1780917310; c=relaxed/simple; bh=NT9SjpeSaUaPs6mTzZoOuRZvfOJboQptqNnZ7k1R9SE=; h=Message-ID:Date:MIME-Version:Subject:To:Cc:References:From: In-Reply-To:Content-Type; b=cOqLAsbznEyVfWlikQfenXw5jIqR2fTC4dGywv1oIIJFKi6/QNM499gZqtV8yHX+gcY/2tOzdZuA4qKe5y/R7mx/OgNBUInItVJ/48NpjsbYFMIZGZFFj8tfv8CUNy2L3OBr5qdOqJCLwLuD3E4lyGTz9dj849UlUCG8jFE2JJ8= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=arm.com; spf=pass smtp.mailfrom=arm.com; dkim=pass (1024-bit key) header.d=arm.com header.i=@arm.com header.b=MtCboeBz; arc=none smtp.client-ip=217.140.110.172 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=arm.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=arm.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=arm.com header.i=@arm.com header.b="MtCboeBz" Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id 1EE8E1C01; Mon, 8 Jun 2026 04:15:03 -0700 (PDT) Received: from [10.57.28.93] (unknown [10.57.28.93]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id 45A3E3F86F; Mon, 8 Jun 2026 04:15:03 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=simple/simple; d=arm.com; s=foss; t=1780917307; bh=NT9SjpeSaUaPs6mTzZoOuRZvfOJboQptqNnZ7k1R9SE=; h=Date:Subject:To:Cc:References:From:In-Reply-To:From; b=MtCboeBzk2dAdFydLOFiNhb5EdBEEhdtHveoclXm8AKsDlHbycpAmJ59UFP3FGlhh Ivseopd9E0d4YjveXKT6AvAyvQLVqub98b68CZwu3SNhMIpz58qZdaOQ1NYTky+TRI uXM0L3c2EL5eG4QbslgeJ0pnOPt/ifx/77lB1ilM= Message-ID: Date: Mon, 8 Jun 2026 12:15:01 +0100 Precedence: bulk X-Mailing-List: kvmarm@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 User-Agent: Mozilla Thunderbird Subject: Re: [PATCH v14 32/44] KVM: arm64: Handle Realm PSCI requests To: Gavin Shan , kvm@vger.kernel.org, kvmarm@lists.linux.dev Cc: Catalin Marinas , Marc Zyngier , Will Deacon , James Morse , Oliver Upton , Suzuki K Poulose , Zenghui Yu , linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, Joey Gouly , Alexandru Elisei , Christoffer Dall , Fuad Tabba , linux-coco@lists.linux.dev, Ganapatrao Kulkarni , Shanker Donthineni , Alper Gun , "Aneesh Kumar K . V" , Emi Kisanuki , Vishal Annapurve , WeiLin.Chang@arm.com, Lorenzo.Pieralisi2@arm.com References: <20260513131757.116630-1-steven.price@arm.com> <20260513131757.116630-33-steven.price@arm.com> <775a0d29-4d92-4ecc-96dd-5b0eaeff1528@redhat.com> From: Steven Price Content-Language: en-GB In-Reply-To: <775a0d29-4d92-4ecc-96dd-5b0eaeff1528@redhat.com> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit On 28/05/2026 07:55, Gavin Shan wrote: > Hi Steve, > > On 5/13/26 11:17 PM, Steven Price wrote: >> The RMM needs to be informed of the target REC when a PSCI call is made >> with an MPIDR argument. >> >> This requirement will be removed in a future release of the RMM 2.0 >> specification but is still required for v2.0-bet1. >> >> Co-developed-by: Suzuki K Poulose >> Signed-off-by: Suzuki K Poulose >> Signed-off-by: Steven Price >> --- >> Chanegs since v13: >>   * The ioctl KVM_ARM_VCPU_RMI_PSCI_COMPLETE has gone. The RMI call is >>     made automatically just before entering the REC again. >> Changes since v12: >>   * Chance return code for non-realms to -ENXIO to better represent that >>     the ioctl is invalid for non-realms (checkpatch is insistent that >>     "ENOSYS means 'invalid syscall nr' and nothing else"). >> Changes since v11: >>   * RMM->RMI renaming. >> Changes since v6: >>   * Use vcpu_is_rec() rather than kvm_is_realm(vcpu->kvm). >>   * Minor renaming/formatting fixes. >> --- >>   arch/arm64/include/asm/kvm_rmi.h |  3 ++ >>   arch/arm64/kvm/psci.c            | 15 ++++++++- >>   arch/arm64/kvm/rmi.c             | 58 ++++++++++++++++++++++++++++++++ >>   3 files changed, 75 insertions(+), 1 deletion(-) >> >> diff --git a/arch/arm64/include/asm/kvm_rmi.h b/arch/arm64/include/ >> asm/kvm_rmi.h >> index b65cfec10dee..eacf82a7467d 100644 >> --- a/arch/arm64/include/asm/kvm_rmi.h >> +++ b/arch/arm64/include/asm/kvm_rmi.h >> @@ -109,6 +109,9 @@ int realm_map_non_secure(struct realm *realm, >>                unsigned long size, >>                enum kvm_pgtable_prot prot, >>                struct kvm_mmu_memory_cache *memcache); >> +int realm_psci_complete(struct kvm_vcpu *source, >> +            struct kvm_vcpu *target, >> +            unsigned long status); >>     static inline bool kvm_realm_is_private_address(struct realm *realm, >>                           unsigned long addr) >> diff --git a/arch/arm64/kvm/psci.c b/arch/arm64/kvm/psci.c >> index 3b5dbe9a0a0e..a2cd55dc7b5b 100644 >> --- a/arch/arm64/kvm/psci.c >> +++ b/arch/arm64/kvm/psci.c >> @@ -103,7 +103,6 @@ static unsigned long kvm_psci_vcpu_on(struct >> kvm_vcpu *source_vcpu) >>         reset_state->reset = true; >>       kvm_make_request(KVM_REQ_VCPU_RESET, vcpu); >> - > > This change isn't supposed to be part of this patch :-) Whoops - indeed it isn't! >>       /* >>        * Make sure the reset request is observed if the RUNNABLE >> mp_state is >>        * observed. >> @@ -142,6 +141,20 @@ static unsigned long >> kvm_psci_vcpu_affinity_info(struct kvm_vcpu *vcpu) >>       /* Ignore other bits of target affinity */ >>       target_affinity &= target_affinity_mask; >>   +    if (vcpu_is_rec(vcpu)) { >> +        struct kvm_vcpu *target_vcpu; >> + >> +        /* RMM supports only zero affinity level */ >> +        if (lowest_affinity_level != 0) >> +            return PSCI_RET_INVALID_PARAMS; >> + >> +        target_vcpu = kvm_mpidr_to_vcpu(kvm, target_affinity); >> +        if (!target_vcpu) >> +            return PSCI_RET_INVALID_PARAMS; >> + >> +        return PSCI_RET_SUCCESS; >> +    } >> + >>       /* >>        * If one or more VCPU matching target affinity are running >>        * then ON else OFF >> diff --git a/arch/arm64/kvm/rmi.c b/arch/arm64/kvm/rmi.c >> index 761b38a4071c..2b03e962ee41 100644 >> --- a/arch/arm64/kvm/rmi.c >> +++ b/arch/arm64/kvm/rmi.c >> @@ -3,6 +3,7 @@ >>    * Copyright (C) 2023-2025 ARM Ltd. >>    */ >>   +#include >>   #include >>     #include >> @@ -127,6 +128,25 @@ static void free_rtt(phys_addr_t phys) >>       kvm_account_pgtable_pages(phys_to_virt(phys), -1); >>   } >>   +int realm_psci_complete(struct kvm_vcpu *source, struct kvm_vcpu >> *target, >> +            unsigned long status) >> +{ >> +    int ret; >> + >> +    /* >> +     * XXX: RMM-v2.0 doesn't require the target REC address for >> completing >> +     * PSCI requests. Temporary hack until RMM implementation catches up >> +     * to the full spec. >> +     */ >> +    ret = rmi_psci_complete(virt_to_phys(source->arch.rec.rec_page), >> +                virt_to_phys(target->arch.rec.rec_page), >> +                status); >> +    if (ret) >> +        return -EINVAL; > >         return -ENXIO; Ack, although as the comment says this should be going away. Thanks, Steve >> + >> +    return 0; >> +} >> + >>   static int realm_rtt_create(struct realm *realm, >>                   unsigned long addr, >>                   int level, >> @@ -1004,6 +1024,41 @@ static void kvm_complete_ripas_change(struct >> kvm_vcpu *vcpu) >>       rec->run->exit.ripas_base = base; >>   } >>   +static void kvm_rec_complete_psci(struct kvm_vcpu *vcpu) >> +{ >> +    struct rec_run *run = vcpu->arch.rec.run; >> +    unsigned long status = PSCI_RET_DENIED; >> +    unsigned long ret = vcpu_get_reg(vcpu, 0); >> +    struct kvm_vcpu *target; >> + >> +    switch (run->exit.gprs[0]) { >> +    /* >> +     * XXX: RMM-v2.0 doesn't cause RMI_EXIT_PSCI for AFFINITY_INFO >> +     * Temporary hack until tf-RMM gets the REC to MPIDR mapping via >> +     * RD Auxiliary granules. >> +     * For now always report SUCCESS >> +     */ >> +    case PSCI_0_2_FN64_AFFINITY_INFO: >> +        status = PSCI_RET_SUCCESS; >> +        break; >> +    case PSCI_0_2_FN64_CPU_ON: { >> +        if (ret != PSCI_RET_SUCCESS && >> +            ret != PSCI_RET_ALREADY_ON) >> +            status = PSCI_RET_DENIED; >> +        else >> +            status = PSCI_RET_SUCCESS; >> +        break; >> +    } >> +    default: >> +        return; >> +    } >> + >> +    target = kvm_mpidr_to_vcpu(vcpu->kvm, run->exit.gprs[1]); >> +    /* RMM makes sure that we don't get RMI_EXIT_PSCI for invalid >> mpidrs */ >> +    if (target) >> +        realm_psci_complete(vcpu, target, status); >> +} >> + >>   /* >>    * kvm_rec_pre_enter - Complete operations before entering a REC >>    * >> @@ -1028,6 +1083,9 @@ int kvm_rec_pre_enter(struct kvm_vcpu *vcpu) >>           for (int i = 0; i < REC_RUN_GPRS; i++) >>               rec->run->enter.gprs[i] = vcpu_get_reg(vcpu, i); >>           break; >> +    case RMI_EXIT_PSCI: >> +        kvm_rec_complete_psci(vcpu); >> +        break; >>       case RMI_EXIT_RIPAS_CHANGE: >>           kvm_complete_ripas_change(vcpu); >>           break; > > Thanks, > Gavin >