From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id C3D66C4167B for ; Tue, 7 Nov 2023 23:36:28 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1344500AbjKGXg3 (ORCPT ); Tue, 7 Nov 2023 18:36:29 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:54480 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S235459AbjKGXgO (ORCPT ); Tue, 7 Nov 2023 18:36:14 -0500 Received: from casper.infradead.org (casper.infradead.org [IPv6:2001:8b0:10b:1236::1]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 16E793A86; Tue, 7 Nov 2023 15:25:05 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=infradead.org; s=casper.20170209; h=MIME-Version:Content-Type:References: In-Reply-To:Date:Cc:To:From:Subject:Message-ID:Sender:Reply-To: Content-Transfer-Encoding:Content-ID:Content-Description; bh=562c+kTyqebhqIjinHXK+0weA3qfzEQhDJM+cW39egE=; b=nvwHpY7ENBkv3+Cww/r+MyLyTT CbB0HeDZ6ZsT3XoXHDtuZJHk6EG7RY1n3FsE869TSCZxBSVni5unHQo7dcdXS0sNFsv2XiOFlzULC hTOdEykrGdHX/ARuMZ+9uuHlURvqiQsDtKLrvKKZE8v4xMAn8PrkyDMNQe+OSmqQR5csmKa5Tke0A FMhQxYh7A98n2kzpf/BI++i3IBm9s4H2dFW7EVeQERQNO5l6pKmgtmiJevWvxzjp52aCJz2uFrSEG uO0FKBkzL0dHwDmaeETE/Xuz3oYfivoW/NO+mLIx6oRk1OeDdmINaycS0SEZvJzT5UfT4jgQykKLI Q58n+P1g==; Received: from [2001:8b0:10b:5:d099:ce97:b761:5eff] (helo=u3832b3a9db3152.ant.amazon.com) by casper.infradead.org with esmtpsa (Exim 4.94.2 #2 (Red Hat Linux)) id 1r0VRI-00FAY8-6H; Tue, 07 Nov 2023 23:24:57 +0000 Message-ID: <37225cb2ab45c842275c2b5b5d84d1bb514a8640.camel@infradead.org> Subject: Re: [PATCH v2] KVM: x86/xen: improve accuracy of Xen timers From: David Woodhouse To: Dongli Zhang , kvm@vger.kernel.org, linux-kernel@vger.kernel.org Cc: Paul Durrant , Sean Christopherson , Paolo Bonzini , Thomas Gleixner , Ingo Molnar , Borislav Petkov , Dave Hansen , x86@kernel.org, "H. Peter Anvin" Date: Tue, 07 Nov 2023 23:24:55 +0000 In-Reply-To: <19f8de0a-17f7-1a25-f2e9-adbf00ecb035@oracle.com> References: <96da7273adfff2a346de9a4a27ce064f6fe0d0a1.camel@infradead.org> <74f32bfae7243a78d0e74b1ba3a2d1ea4a4a7518.camel@infradead.org> <2bd5d543-08a0-a0f6-0f59-b8724a2d8d75@oracle.com> <12e8ade22fe6c1e6bec74e60e8213302a7da635e.camel@infradead.org> <19f8de0a-17f7-1a25-f2e9-adbf00ecb035@oracle.com> Content-Type: multipart/signed; micalg="sha-256"; protocol="application/pkcs7-signature"; boundary="=-fuxoxWpgbdV2bQ6nS6ko" User-Agent: Evolution 3.44.4-0ubuntu2 MIME-Version: 1.0 X-SRS-Rewrite: SMTP reverse-path rewritten from by casper.infradead.org. See http://www.infradead.org/rpr.html Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org --=-fuxoxWpgbdV2bQ6nS6ko Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Tue, 2023-11-07 at 15:07 -0800, Dongli Zhang wrote: > Thank you very much for the detailed explanation. >=20 > I agree it is important to resolve the "now" problem. I guess the KVM lap= ic > deadline timer has the "now" problem as well. I think so. And quite gratuitously so, since it just does: now =3D ktime_get(); guest_tsc =3D kvm_read_l1_tsc(vcpu, rdtsc()); Couldn't that trivially be changed to kvm_get_monotonic_and_clockread()? Thankfully, it's defined in the time domain of the guest TSC, not the kvmclock, so it doesn't suffer the same drift issue as the Xen timer. > I just notice my question missed a key prerequisite: >=20 > Would you mind helping explain the time domain of the "oneshot.timeout_ab= s_ns"? >=20 > While it is the absolute nanosecond value at the VM side, on which time d= omain > it is based? It's the kvmclock. Xen offers as Xen PV clock to its guests using *precisely* the same pvclock structure as KVM does.=20 > 1. Is oneshot.timeout_abs_ns based on the xen pvclock (freq=3DNSEC_PER_SE= C)? >=20 > 2. Is oneshot.timeout_abs_ns based on tsc from VM side? >=20 > 3. Is oneshot.timeout_abs_ns based on monotonic/raw clock at VM side? >=20 > 4. Or it is based on wallclock? >=20 > I think the OS does not have a concept of nanoseconds. It is derived from= a > clocksource. It's the kvmclock. The guest derives it from the guest TSC using the pvclock information (mul/shift/offset) that KVM provides to the guest. The kvm_setup_guest_pvclock() function is potentially called *three* times from kvm_guest_time_update(). Once for the KVM pv time MSR, once for the pvclock structure in the Xen vcpu_info, and finally for the pvclock structure which Xen makes available to userspace for vDSO timekeeping. > If it is based on pvclock, is it based on the pvclock from a specific vCP= U, as > both pvclock and timer are per-vCPU. Yes, it is per-vCPU. Although in the sane case the TSCs on all vCPUs will match and the mul/shift/offset provided by KVM won't actually differ. Even in the insane case where guest TSCs are out of sync, surely the pvclock information will differ only in order to ensure that the *result* in nanoseconds does not? I conveniently ducked this question in my patch by only supporting the CONSTANT_TSC case, and not the case where we happen to know the (potentially different) TSC frequencies on all the different pCPUs and vCPUs. >=20 > E.g., according to the KVM lapic deadline timer, all values are based on = (1) the > tsc value, (2)on the current vCPU. >=20 >=20 > 1949 static void start_sw_tscdeadline(struct kvm_lapic *apic) > 1950 { > 1951=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct kvm_timer *kt= imer =3D &apic->lapic_timer; > 1952=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 u64 guest_tsc, tscde= adline =3D ktimer->tscdeadline; > 1953=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 u64 ns =3D 0; > 1954=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 ktime_t expire; > 1955=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct kvm_vcpu *vcp= u =3D apic->vcpu; > 1956=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 unsigned long this_t= sc_khz =3D vcpu->arch.virtual_tsc_khz; > 1957=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 unsigned long flags; > 1958=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 ktime_t now; > 1959 > 1960=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (unlikely(!tscdea= dline || !this_tsc_khz)) > 1961=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2= =A0=C2=A0=C2=A0=C2=A0=C2=A0 return; > 1962 > 1963=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 local_irq_save(flags= ); > 1964 > 1965=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 now =3D ktime_get(); > 1966=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 guest_tsc =3D kvm_re= ad_l1_tsc(vcpu, rdtsc()); > 1967 > 1968=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 ns =3D (tscdeadline = - guest_tsc) * 1000000ULL; > 1969=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 do_div(ns, this_tsc_= khz); >=20 >=20 > Sorry if I make the question very confusing. The core question is: where = and > from which clocksource the abs nanosecond value is from? What will happen= if the > Xen VM uses HPET as clocksource, while xen timer as clock event? If the guest uses HPET as clocksource and Xen timer as clockevents, then keeping itself in sync is the *guest's* problem. The Xen timer is defined in terms of nanoseconds since guest start, as provided in the pvclock information described above. Hope that helps! --=-fuxoxWpgbdV2bQ6nS6ko Content-Type: application/pkcs7-signature; name="smime.p7s" Content-Disposition: attachment; filename="smime.p7s" Content-Transfer-Encoding: base64 MIAGCSqGSIb3DQEHAqCAMIACAQExDzANBglghkgBZQMEAgEFADCABgkqhkiG9w0BBwEAAKCCEkQw ggYQMIID+KADAgECAhBNlCwQ1DvglAnFgS06KwZPMA0GCSqGSIb3DQEBDAUAMIGIMQswCQYDVQQG EwJVUzETMBEGA1UECBMKTmV3IEplcnNleTEUMBIGA1UEBxMLSmVyc2V5IENpdHkxHjAcBgNVBAoT FVRoZSBVU0VSVFJVU1QgTmV0d29yazEuMCwGA1UEAxMlVVNFUlRydXN0IFJTQSBDZXJ0aWZpY2F0 aW9uIEF1dGhvcml0eTAeFw0xODExMDIwMDAwMDBaFw0zMDEyMzEyMzU5NTlaMIGWMQswCQYDVQQG EwJHQjEbMBkGA1UECBMSR3JlYXRlciBNYW5jaGVzdGVyMRAwDgYDVQQHEwdTYWxmb3JkMRgwFgYD VQQKEw9TZWN0aWdvIExpbWl0ZWQxPjA8BgNVBAMTNVNlY3RpZ28gUlNBIENsaWVudCBBdXRoZW50 aWNhdGlvbiBhbmQgU2VjdXJlIEVtYWlsIENBMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKC AQEAyjztlApB/975Rrno1jvm2pK/KxBOqhq8gr2+JhwpKirSzZxQgT9tlC7zl6hn1fXjSo5MqXUf ItMltrMaXqcESJuK8dtK56NCSrq4iDKaKq9NxOXFmqXX2zN8HHGjQ2b2Xv0v1L5Nk1MQPKA19xeW QcpGEGFUUd0kN+oHox+L9aV1rjfNiCj3bJk6kJaOPabPi2503nn/ITX5e8WfPnGw4VuZ79Khj1YB rf24k5Ee1sLTHsLtpiK9OjG4iQRBdq6Z/TlVx/hGAez5h36bBJMxqdHLpdwIUkTqT8se3ed0PewD ch/8kHPo5fZl5u1B0ecpq/sDN/5sCG52Ds+QU5O5EwIDAQABo4IBZDCCAWAwHwYDVR0jBBgwFoAU U3m/WqorSs9UgOHYm8Cd8rIDZsswHQYDVR0OBBYEFAnA8vwL2pTbX/4r36iZQs/J4K0AMA4GA1Ud DwEB/wQEAwIBhjASBgNVHRMBAf8ECDAGAQH/AgEAMB0GA1UdJQQWMBQGCCsGAQUFBwMCBggrBgEF BQcDBDARBgNVHSAECjAIMAYGBFUdIAAwUAYDVR0fBEkwRzBFoEOgQYY/aHR0cDovL2NybC51c2Vy dHJ1c3QuY29tL1VTRVJUcnVzdFJTQUNlcnRpZmljYXRpb25BdXRob3JpdHkuY3JsMHYGCCsGAQUF BwEBBGowaDA/BggrBgEFBQcwAoYzaHR0cDovL2NydC51c2VydHJ1c3QuY29tL1VTRVJUcnVzdFJT QUFkZFRydXN0Q0EuY3J0MCUGCCsGAQUFBzABhhlodHRwOi8vb2NzcC51c2VydHJ1c3QuY29tMA0G CSqGSIb3DQEBDAUAA4ICAQBBRHUAqznCFfXejpVtMnFojADdF9d6HBA4kMjjsb0XMZHztuOCtKF+ xswhh2GqkW5JQrM8zVlU+A2VP72Ky2nlRA1GwmIPgou74TZ/XTarHG8zdMSgaDrkVYzz1g3nIVO9 IHk96VwsacIvBF8JfqIs+8aWH2PfSUrNxP6Ys7U0sZYx4rXD6+cqFq/ZW5BUfClN/rhk2ddQXyn7 kkmka2RQb9d90nmNHdgKrwfQ49mQ2hWQNDkJJIXwKjYA6VUR/fZUFeCUisdDe/0ABLTI+jheXUV1 eoYV7lNwNBKpeHdNuO6Aacb533JlfeUHxvBz9OfYWUiXu09sMAviM11Q0DuMZ5760CdO2VnpsXP4 KxaYIhvqPqUMWqRdWyn7crItNkZeroXaecG03i3mM7dkiPaCkgocBg0EBYsbZDZ8bsG3a08LwEsL 1Ygz3SBsyECa0waq4hOf/Z85F2w2ZpXfP+w8q4ifwO90SGZZV+HR/Jh6rEaVPDRF/CEGVqR1hiuQ OZ1YL5ezMTX0ZSLwrymUE0pwi/KDaiYB15uswgeIAcA6JzPFf9pLkAFFWs1QNyN++niFhsM47qod x/PL+5jR87myx5uYdBEQkkDc+lKB1Wct6ucXqm2EmsaQ0M95QjTmy+rDWjkDYdw3Ms6mSWE3Bn7i 5ZgtwCLXgAIe5W8mybM2JzCCBhQwggT8oAMCAQICEQDGvhmWZ0DEAx0oURL6O6l+MA0GCSqGSIb3 DQEBCwUAMIGWMQswCQYDVQQGEwJHQjEbMBkGA1UECBMSR3JlYXRlciBNYW5jaGVzdGVyMRAwDgYD VQQHEwdTYWxmb3JkMRgwFgYDVQQKEw9TZWN0aWdvIExpbWl0ZWQxPjA8BgNVBAMTNVNlY3RpZ28g UlNBIENsaWVudCBBdXRoZW50aWNhdGlvbiBhbmQgU2VjdXJlIEVtYWlsIENBMB4XDTIyMDEwNzAw MDAwMFoXDTI1MDEwNjIzNTk1OVowJDEiMCAGCSqGSIb3DQEJARYTZHdtdzJAaW5mcmFkZWFkLm9y ZzCCAiIwDQYJKoZIhvcNAQEBBQADggIPADCCAgoCggIBALQ3GpC2bomUqk+91wLYBzDMcCj5C9m6 oZaHwvmIdXftOgTbCJXADo6G9T7BBAebw2JV38EINgKpy/ZHh7htyAkWYVoFsFPrwHounto8xTsy SSePMiPlmIdQ10BcVSXMUJ3Juu16GlWOnAMJY2oYfEzmE7uT9YgcBqKCo65pTFmOnR/VVbjJk4K2 xE34GC2nAdUQkPFuyaFisicc6HRMOYXPuF0DuwITEKnjxgNjP+qDrh0db7PAjO1D4d5ftfrsf+kd RR4gKVGSk8Tz2WwvtLAroJM4nXjNPIBJNT4w/FWWc/5qPHJy2U+eITZ5LLE5s45mX2oPFknWqxBo bQZ8a9dsZ3dSPZBvE9ZrmtFLrVrN4eo1jsXgAp1+p7bkfqd3BgBEmfsYWlBXO8rVXfvPgLs32VdV NZxb/CDWPqBsiYv0Hv3HPsz07j5b+/cVoWqyHDKzkaVbxfq/7auNVRmPB3v5SWEsH8xi4Bez2V9U KxfYCnqsjp8RaC2/khxKt0A552Eaxnz/4ly/2C7wkwTQnBmdlFYhAflWKQ03Ufiu8t3iBE3VJbc2 5oMrglj7TRZrmKq3CkbFnX0fyulB+kHimrt6PIWn7kgyl9aelIl6vtbhMA+l0nfrsORMa4kobqQ5 C5rveVgmcIad67EDa+UqEKy/GltUwlSh6xy+TrK1tzDvAgMBAAGjggHMMIIByDAfBgNVHSMEGDAW gBQJwPL8C9qU21/+K9+omULPyeCtADAdBgNVHQ4EFgQUzMeDMcimo0oz8o1R1Nver3ZVpSkwDgYD VR0PAQH/BAQDAgWgMAwGA1UdEwEB/wQCMAAwHQYDVR0lBBYwFAYIKwYBBQUHAwQGCCsGAQUFBwMC MEAGA1UdIAQ5MDcwNQYMKwYBBAGyMQECAQEBMCUwIwYIKwYBBQUHAgEWF2h0dHBzOi8vc2VjdGln by5jb20vQ1BTMFoGA1UdHwRTMFEwT6BNoEuGSWh0dHA6Ly9jcmwuc2VjdGlnby5jb20vU2VjdGln b1JTQUNsaWVudEF1dGhlbnRpY2F0aW9uYW5kU2VjdXJlRW1haWxDQS5jcmwwgYoGCCsGAQUFBwEB BH4wfDBVBggrBgEFBQcwAoZJaHR0cDovL2NydC5zZWN0aWdvLmNvbS9TZWN0aWdvUlNBQ2xpZW50 QXV0aGVudGljYXRpb25hbmRTZWN1cmVFbWFpbENBLmNydDAjBggrBgEFBQcwAYYXaHR0cDovL29j c3Auc2VjdGlnby5jb20wHgYDVR0RBBcwFYETZHdtdzJAaW5mcmFkZWFkLm9yZzANBgkqhkiG9w0B AQsFAAOCAQEAyW6MUir5dm495teKqAQjDJwuFCi35h4xgnQvQ/fzPXmtR9t54rpmI2TfyvcKgOXp qa7BGXNFfh1JsqexVkIqZP9uWB2J+uVMD+XZEs/KYNNX2PvIlSPrzIB4Z2wyIGQpaPLlYflrrVFK v9CjT2zdqvy2maK7HKOQRt3BiJbVG5lRiwbbygldcALEV9ChWFfgSXvrWDZspnU3Gjw/rMHrGnql Htlyebp3pf3fSS9kzQ1FVtVIDrL6eqhTwJxe+pXSMMqFiN0whpBtXdyDjzBtQTaZJ7zTT/vlehc/ tDuqZwGHm/YJy883Ll+GP3NvOkgaRGWEuYWJJ6hFCkXYjyR9IzCCBhQwggT8oAMCAQICEQDGvhmW Z0DEAx0oURL6O6l+MA0GCSqGSIb3DQEBCwUAMIGWMQswCQYDVQQGEwJHQjEbMBkGA1UECBMSR3Jl YXRlciBNYW5jaGVzdGVyMRAwDgYDVQQHEwdTYWxmb3JkMRgwFgYDVQQKEw9TZWN0aWdvIExpbWl0 ZWQxPjA8BgNVBAMTNVNlY3RpZ28gUlNBIENsaWVudCBBdXRoZW50aWNhdGlvbiBhbmQgU2VjdXJl IEVtYWlsIENBMB4XDTIyMDEwNzAwMDAwMFoXDTI1MDEwNjIzNTk1OVowJDEiMCAGCSqGSIb3DQEJ ARYTZHdtdzJAaW5mcmFkZWFkLm9yZzCCAiIwDQYJKoZIhvcNAQEBBQADggIPADCCAgoCggIBALQ3 GpC2bomUqk+91wLYBzDMcCj5C9m6oZaHwvmIdXftOgTbCJXADo6G9T7BBAebw2JV38EINgKpy/ZH h7htyAkWYVoFsFPrwHounto8xTsySSePMiPlmIdQ10BcVSXMUJ3Juu16GlWOnAMJY2oYfEzmE7uT 9YgcBqKCo65pTFmOnR/VVbjJk4K2xE34GC2nAdUQkPFuyaFisicc6HRMOYXPuF0DuwITEKnjxgNj P+qDrh0db7PAjO1D4d5ftfrsf+kdRR4gKVGSk8Tz2WwvtLAroJM4nXjNPIBJNT4w/FWWc/5qPHJy 2U+eITZ5LLE5s45mX2oPFknWqxBobQZ8a9dsZ3dSPZBvE9ZrmtFLrVrN4eo1jsXgAp1+p7bkfqd3 BgBEmfsYWlBXO8rVXfvPgLs32VdVNZxb/CDWPqBsiYv0Hv3HPsz07j5b+/cVoWqyHDKzkaVbxfq/ 7auNVRmPB3v5SWEsH8xi4Bez2V9UKxfYCnqsjp8RaC2/khxKt0A552Eaxnz/4ly/2C7wkwTQnBmd lFYhAflWKQ03Ufiu8t3iBE3VJbc25oMrglj7TRZrmKq3CkbFnX0fyulB+kHimrt6PIWn7kgyl9ae lIl6vtbhMA+l0nfrsORMa4kobqQ5C5rveVgmcIad67EDa+UqEKy/GltUwlSh6xy+TrK1tzDvAgMB AAGjggHMMIIByDAfBgNVHSMEGDAWgBQJwPL8C9qU21/+K9+omULPyeCtADAdBgNVHQ4EFgQUzMeD Mcimo0oz8o1R1Nver3ZVpSkwDgYDVR0PAQH/BAQDAgWgMAwGA1UdEwEB/wQCMAAwHQYDVR0lBBYw FAYIKwYBBQUHAwQGCCsGAQUFBwMCMEAGA1UdIAQ5MDcwNQYMKwYBBAGyMQECAQEBMCUwIwYIKwYB BQUHAgEWF2h0dHBzOi8vc2VjdGlnby5jb20vQ1BTMFoGA1UdHwRTMFEwT6BNoEuGSWh0dHA6Ly9j cmwuc2VjdGlnby5jb20vU2VjdGlnb1JTQUNsaWVudEF1dGhlbnRpY2F0aW9uYW5kU2VjdXJlRW1h aWxDQS5jcmwwgYoGCCsGAQUFBwEBBH4wfDBVBggrBgEFBQcwAoZJaHR0cDovL2NydC5zZWN0aWdv LmNvbS9TZWN0aWdvUlNBQ2xpZW50QXV0aGVudGljYXRpb25hbmRTZWN1cmVFbWFpbENBLmNydDAj BggrBgEFBQcwAYYXaHR0cDovL29jc3Auc2VjdGlnby5jb20wHgYDVR0RBBcwFYETZHdtdzJAaW5m cmFkZWFkLm9yZzANBgkqhkiG9w0BAQsFAAOCAQEAyW6MUir5dm495teKqAQjDJwuFCi35h4xgnQv Q/fzPXmtR9t54rpmI2TfyvcKgOXpqa7BGXNFfh1JsqexVkIqZP9uWB2J+uVMD+XZEs/KYNNX2PvI lSPrzIB4Z2wyIGQpaPLlYflrrVFKv9CjT2zdqvy2maK7HKOQRt3BiJbVG5lRiwbbygldcALEV9Ch WFfgSXvrWDZspnU3Gjw/rMHrGnqlHtlyebp3pf3fSS9kzQ1FVtVIDrL6eqhTwJxe+pXSMMqFiN0w hpBtXdyDjzBtQTaZJ7zTT/vlehc/tDuqZwGHm/YJy883Ll+GP3NvOkgaRGWEuYWJJ6hFCkXYjyR9 IzGCBMcwggTDAgEBMIGsMIGWMQswCQYDVQQGEwJHQjEbMBkGA1UECBMSR3JlYXRlciBNYW5jaGVz dGVyMRAwDgYDVQQHEwdTYWxmb3JkMRgwFgYDVQQKEw9TZWN0aWdvIExpbWl0ZWQxPjA8BgNVBAMT NVNlY3RpZ28gUlNBIENsaWVudCBBdXRoZW50aWNhdGlvbiBhbmQgU2VjdXJlIEVtYWlsIENBAhEA xr4ZlmdAxAMdKFES+jupfjANBglghkgBZQMEAgEFAKCCAeswGAYJKoZIhvcNAQkDMQsGCSqGSIb3 DQEHATAcBgkqhkiG9w0BCQUxDxcNMjMxMTA3MjMyNDU1WjAvBgkqhkiG9w0BCQQxIgQgGEwoImXI eqwQOgnZlU6v+MBi5FAxAbXixL054mF66m8wgb0GCSsGAQQBgjcQBDGBrzCBrDCBljELMAkGA1UE BhMCR0IxGzAZBgNVBAgTEkdyZWF0ZXIgTWFuY2hlc3RlcjEQMA4GA1UEBxMHU2FsZm9yZDEYMBYG A1UEChMPU2VjdGlnbyBMaW1pdGVkMT4wPAYDVQQDEzVTZWN0aWdvIFJTQSBDbGllbnQgQXV0aGVu dGljYXRpb24gYW5kIFNlY3VyZSBFbWFpbCBDQQIRAMa+GZZnQMQDHShREvo7qX4wgb8GCyqGSIb3 DQEJEAILMYGvoIGsMIGWMQswCQYDVQQGEwJHQjEbMBkGA1UECBMSR3JlYXRlciBNYW5jaGVzdGVy MRAwDgYDVQQHEwdTYWxmb3JkMRgwFgYDVQQKEw9TZWN0aWdvIExpbWl0ZWQxPjA8BgNVBAMTNVNl Y3RpZ28gUlNBIENsaWVudCBBdXRoZW50aWNhdGlvbiBhbmQgU2VjdXJlIEVtYWlsIENBAhEAxr4Z lmdAxAMdKFES+jupfjANBgkqhkiG9w0BAQEFAASCAgB7frAiURz3P6gZTtBgixTB73cYgByTRJ9a zHIqyFaSVmY0F11qqrZ+XnynKmorJxkbmPq0UxTrQblvq2sgIpwzWsmoryZ1Wy3ONl5DJNb4c0NP ZfvrL8c4Cxti6F1W8W1b6bomnhS9QDErLFhvjjYaFA/ByFweHXXJcZwn1FBM+2KYKKyQK4xCI6n6 E9NE02+RIai0IPL5xEAbp7M/zDahGM6DBZ3HB+gXvLTHOdIKO2GKqavs4THiiR1nK+J8KI/U62Dk AmFF1uIhuTLOrSq7Uo4Sua+7CWv3ZPXyNMYG2TzsBLzNoYTrBLFKchH4AqjxPNl5+FEP6rfQ+gQ+ JKKdclD2ypYPZ2f6wFQVnllSY6mG4a3VcxTM3nPi9WdS4V8dhMQNd2T68yMisA9Bqy3sE8pR0Il+ 31ryBzxvdgS4eAgtHR9cAdLRR+NVCkr1XuZcLPfBvJq3bv7KH3AyOf5AY57aL9xoRLQnpigYRcCI ZjAvcSVHjAG3kfU6D3XrJs8uZvHZzRQhuswCqRO82sDmcs3mPg6qXvqOdFGx6cTh0sz+FOe+7FrE nXilFNjzmlZ5/JO49hoR+dFXaQxZqQzJ1HjxcwwtKn/8uhIu8W4pSGEYTHcKun5NdJ6UK/u3Ustx tm27RZGAP1RDJzjpmBM/TDEIqDEqtbehlhx0zACiawAAAAAAAA== --=-fuxoxWpgbdV2bQ6nS6ko--